Carbonatite and silicate melt metasomatism of depleted mantle surrounding the Hawaiian plume: origin...

Carbonatite and silicate melt metasomatism of the mantlesurrounding the Hawaiian plume: Evidence from volatiles,trace elements, and radiogenic isotopes in rejuvenated-stagelavas from Niihau, Hawaii

Jacqueline DixonDepartment of Geological Sciences, University of Miami, P.O. Box 249176, Coral Gables, Florida 33124-9176, USA([email protected])

David A. ClagueMonterey Bay Aquarium Research Institute, 7700 Sandholdt Road, Moss Landing, California 95039, USA

Brian CousensOttawa-Carleton Geoscience Centre, Department of Earth Sciences, Carleton University, 1125 Colonel By Drive,Ottawa, Ontario K1S 5B6, Canada

Maria Luisa MonsalveDepartment of Geological Sciences, University of Miami, P.O. Box 249176, Coral Gables, Florida 33124-9176, USA

Jessika UhlOttawa-Carleton Geoscience Centre, Department of Earth Sciences, Carleton University, 1125 Colonel By Drive,Ottawa, Ontario K1S 5B6, Canada

[1] We present new volatile, trace element, and radiogenic isotopic compositions for rejuvenated-stagelavas erupted on Niihau and its submarine northwest flank. Niihau rejuvenated-stage Kiekie Basalt lavasare mildly alkalic and are isotopically similar to, though shifted to higher 87Sr/86Sr and lower 206Pb/204Pbthan, rejuvenated-stage lavas erupted on other islands and marginal seafloor settings. Kiekie lavas displaytrace element heterogeneity greater than that of other rejuvenated-stage lavas, with enrichments in Ba, Sr,and light-rare earth elements resulting in high and highly variable Ba/Th and Sr/Ce. The high Ba/Th lavasare among the least silica-undersaturated of the rejuvenated-stage suite, implying that the greatestenrichments are associated with the largest extents of melting. Kiekie lavas also have high and variableH2O/Ce and Cl/La, up to 620 and 39, respectively. We model the trace element concentrations of mostrejuvenated-stage lavas by small degrees (�1% to 9%) of melting of depleted peridotite recentlymetasomatized by a few percent of an enriched incipient melt (0.5% melting) of the Hawaiian plume.Kiekie lavas are best explained by 4% to 13% partial melting of a peridotite source metasomatized by up to0.2% carbonatite, similar in composition to oceanic carbonatites from the Canary and Cape Verde Islands,with lower proportion of incipient melt than that for other rejuvenated-stage lavas. Primary H2O and Cl ofthe carbonatite component must be high, but variability in the volatile data may be caused by heterogeneityin the carbonatite composition and/or interaction with seawater. Our model is consistent with predictionsbased on carbonated eclogite and peridotite melting experiments in which (1) carbonated eclogite andperidotite within the Hawaiian plume are the first to melt during plume ascent; (2) carbonatite meltmetasomatizes plume and surrounding depleted peridotite; (3) as the plume rises, silica-undersaturatedsilicate melts are also produced and contribute to the metasomatic signature. The metasomatic component

G3G3GeochemistryGeophysics

Geosystems

Published by AGU and the Geochemical Society

AN ELECTRONIC JOURNAL OF THE EARTH SCIENCES

GeochemistryGeophysics

Geosystems

Article

Volume 9, Number 9

27 September 2008

Q09005, doi:10.1029/2008GC002076

ISSN: 1525-2027

Copyright 2008 by the American Geophysical Union 1 of 34

is best preserved at the margins of the plume, where low extents of melting of the metasomatized depletedmantle surrounding the plume are sampled during flexural uplift. Formation of carbonatite melts mayprovide a mechanism to transfer plume He to the margins of the plume.

Components: 21,917 words, 17 figures, 11 tables.

Keywords: Hawaii; alkalic; trace elements; volatiles; metasomatism; rejuvenated-stage lava.

Index Terms: 1037 Geochemistry: Magma genesis and partial melting (3619); 1038 Geochemistry: Mantle processes

(3621); 8410 Volcanology: Geochemical modeling (1009, 3610).

Received 27 April 2008; Revised 24 July 2008; Accepted 29 July 2008; Published 27 September 2008.

Dixon, J., D. A. Clague, B. Cousens, M. L. Monsalve, and J. Uhl (2008), Carbonatite and silicate melt metasomatism of the

mantle surrounding the Hawaiian plume: Evidence from volatiles, trace elements, and radiogenic isotopes in rejuvenated-

stage lavas from Niihau, Hawaii, Geochem. Geophys. Geosyst., 9, Q09005, doi:10.1029/2008GC002076.

1. Introduction

[2] Hawaiian volcanoes, dominantly shields oftholeiitic basalt, form as the Pacific plate migratesover a melting anomaly in the mantle, hereafterreferred to as the Hawaiian plume. As these shieldsmigrate away from the plume, strongly alkaliclavas, forming the rejuvenated-stage of volcanism,may erupt from vents scattered on the tholeiiticshields after an interval of erosion lasting for 0.25–2.5 million years [e.g., Ozawa et al., 2005; Clague,1987, and references therein]. These lavas occur onthe shields of West Maui (Lahaina Basalt), EastMolokai (Kalaupapa Basalt), Koolau (HonoluluVolcanics), Kauai (Koloa Volcanics), and Niihau(Kiekie Basalt), with compositions including alkaliolivine basalt, basanite, nephelinite, and nephelinemelilitite [e.g., Clague and Frey, 1982; Feigenson,1984; Garcia et al., 1986; Chen and Frey, 1983,1985; Clague, 1987; Clague and Dalrymple, 1988;Maaløe et al., 1992; Bergmanis et al., 2000;Fekiacova et al., 2007]. In addition to the subaer-ially erupted, rejuvenated-stage lavas, submarinelavas with similar characteristics have been col-lected on the seafloor surrounding the islands(peripheral lavas) [e.g., Lipman et al., 1989;Clague and Dalrymple, 1989; Clague et al.,1990, 2006; Dixon et al., 1997; Frey et al., 2000;Hanyu et al., 2005], including volcanics eruptedupstream of the plume (preshield or precursory-stage eruptions).

[3] These alkalic lavas are generated by smallextents of partial melting resulting from decom-pression during lithospheric flexural uplift as theoceanic plate (precursory and peripheral) or shield(rejuvenated-stage) passes over the flexural arch,which occurs �200 km from the center of loading

by the growth of the Hawaiian Islands [e.g.,Clague and Dalrymple, 1987; Lipman et al.,1989; ten Brink, 1991; Ribe and Christensen,1999; Ozawa et al., 2005; Bianco et al., 2005].Because precursory, peripheral, and rejuvenated-stage lavas are generated by the same mechanism,we use rejuvenated-stage to refer to lavas eruptedin all three settings.

[4] Radiogenic isotopic compositions of rejuvenated-stage lavas are consistent with mixing between theHawaiian plume and a depleted mantle component[e.g., Yang et al., 2003]. Though a variety of modelshave been proposed [e.g., Chen and Frey, 1983,1985; Sims et al., 1995; Frey et al., 2000; Reiners,2002], the enrichment of incompatible elementscoupled with low 87Sr/86Sr and high 143Nd/144Ndrelative to bulk Earth is best explained by melting ofdepleted mantle recently metasomatized byenriched incipient melt (<2% melting) from theHawaiian plume [Yang et al., 2003]. Recent radio-genic isotopic studies [Bizimis et al., 2005; Frey etal., 2005; Salters et al., 2006; Fekiacova et al.,2007] support this model by showing the depletedcomponent is most likely derived from depletedmantle thermally entrained by the upwellingHawaiian plume and not from the Pacific lithosphere.

[5] In this paper, we present new compositionaldata for rejuvenated-stage lavas from Niihau, theKiekie Basalt, including submarine lavas eruptedon the northwest flank. These lavas provide aunique opportunity to study temporal and spatialvariations in magma generation processes associ-ated with Hawaiian volcanism. In particular, thesubmarine lavas allow us to investigate magmaticvariations with respect to water and carbon diox-ide, which are partially to mostly retained inquenched glassy rinds. We will show that trace

GeochemistryGeophysicsGeosystems G3G3

dixon et al.: volatiles and trace elements in lavas 10.1029/2008GC002076

2 of 34

element compositions of the lavas are bestexplained by the presence of carbonate in theHawaiian plume and the occurrence of carbonatitic,as well as silicate, melt metasomatism of the mantlesource for rejuvenated-stage Hawaiian magmas.

2. Samples

[6] The location of Niihau within the Hawaiianchain is shown in Figure 1. Locations of subaerialKiekie lavas are shown on Figure 2. Subaerialrejuvenated-stage Kiekie Basalt on the island ofNiihau are alkalic basalts erupted 0.3 to 3.5 millionyears ago [Clague and Dalrymple, 1987] and makeup more than half of the subaerial lavas [Stearnsand Macdonald, 1947; Clague, 1987]. Samples ofyoung lava flows and volcanic cones on the sub-marine northwest flank of Niihau Island werecollected during four dives of the remote operatedvehicle (ROV) Tiburon on the Monterey Bay

Aquarium Research Institute (MBARI) R/VWesternFlyer in 2001 (dives T317, T318, T322, and T323;Figure 3). The ages of the Niihau cones are estimatedto be <1 Ma based on the thickness of Mn-oxidecoating on the rocks [Moore and Clague, 2004].

[7] Compositions of the Kiekie lavas will be com-pared to those of other rejuvenated-stage lavas, aswell as to compositions of lavas from variousHawaiian shields. Data sources are listed in Table 1.The data set used is internally consistent; traceelement data are all analyzed in the same labora-tory using ICP-MS. Only samples with MgO >6 wt% are plotted to avoid samples that haveundergone multiphase fractionation. Shield lavasinclude Kilauea Puna Ridge, South Kona Land-slide, Kohala, Molokai, Kahoolawe, Lanai, theWaianae Landslide, Mahukona Terrace, and theNiihau Paniau Shield. Lavas with compositionsrepresentative of the main rejuvenated-stage trend(blue diamonds on all figures) include lavas fromthe North Arch, Kauai-Oahu Channel, West Maui(Lahaina Basalt), East Molokai (Kalaupapa Basalt),and Kauai (Koloa Volcanics). Other rejuvenated-stage lavas highlighted with separate symbols in-clude those from Niihau (this study), the SouthFigure 1. Location map showing Geological Long-

Range Inclined Asdic (GLORIA) acoustic imagery ofthe sea floor around Niihau. Black rectangle in upperright corner shows general location of Hawaiian Islands.Bright areas are young volcanics with high backscatter,in contrast to dark, low-backscatter sediment-covered seafloor. Red rectangles show areas enlarged in Figures 2and 3.

Figure 2. Map of the island of Niihau with topographydisplayed as Sun-illuminated from the southeast. Loca-tions (red circles) and sample numbers of the subaerialvents comprising the Kiekie Volcanics are superposed.


dixon et al.: volatiles and trace elements in lavas 10.1029/2008GC002076dixon et al.: volatiles and trace elements in lavas 10.1029/2008GC002076

3 of 34

Arch, Oahu (Honolulu Volcanics), and Kaula Island(nephelinites).

3. Analytical Techniques

[8] Fresh glasses from rinds on lava were analyzedby electron microprobe with a JEOL Superprobe atthe U.S. Geological Survey in Menlo Park usingnatural and synthetic standards (Table 2). Methodsand analytical precision and accuracy are describedby Davis et al. [1994].

[9] Nineteen whole-rock submarine lava samplesfrom dives T317, T318, T322, and T323 wereanalyzed for major and trace elements using acombination of X-ray fluorescence (XRF) andinductively coupled plasma-mass spectrometry(ICP-MS) techniques at the GeoAnalytical Labo-ratory at Washington State University (Tables 3aand 3b), using techniques described by Knaack etal. [1994]. An additional seven glass samples wereanalyzed by ICP-MS at the GeoAnalytical Labo-ratory (also in Tables 3a and 3b). Major elementanalyses for 26 subaerial Kiekie Basalt sampleswere analyzed at laboratories of the U. S. Geolog-

ical Survey in Denver, Colorado using wet chem-istry [Jackson et al., 1987]. Trace elements for thesubaerial samples were analyzed using a combina-tion of ICP-MS at GeoAnalytical Laboratory andX-ray fluorescence spectrometry (XRF) at theUniversity of Massachusetts.

[10] Concentrations of dissolved water and carbondioxide were measured in submarine glasses using

Table 1. Sources of Data and Group Classification ofHawaiian Shield and Rejuvenated-Stage Lavas

Classification Source

Shield Lavas (GrayCircles on All Figures)

Kilauea Puna RidgeSouth Kona LandslideKohalaMolokaiKahoolaweLanaiWaianae LandslideMahukona TerraceNiihau Paniau Shield

Representative Rejuvenated-Stage Trend (Blue Diamonds on All Figures)

Kauai-Oahu Channel Dixon and Clague [2001];D. Clague et al.(manuscript in preparation, 2008)

West Maui lava LA35; Gaffney et al. [2004]North Arch Clague et al. [1990]

Dixon et al. [1997]Frey et al. [2000]Yang et al. [2003]Hanyu et al. [2005]Davis et al. [1994]

Molokai Clague and Frey [1982]Clague and Moore [2002]Xu et al. [2005]

Koloa Volcanics fromKauai

Clague and Dalrymple [1988]

Maaløe et al. [1992]Reiners and Nelson [1998]

Other Rejuvenated-StageLavas (Separate Symbols)

Niihau this studySouth Arch Lipman et al. [1989]

Sims et al. [1995]Dixon and Clague [2001]Clague et al. (manuscript inpreparation, 2008)

Honolulu Serieson Oahu

Clague and Frey [1982]

Yang et al. [2003]

Kaula nephleinites Garcia et al. [1986]

Figure 3. Locations of dives on the northwest flank ofNiihau by Monterey Bay Aquarium Research Institute(MBARI) R/V Western Flyer and remote operatedvehicle Tiburon in 2001. Bathymetry from MBARIHawaiian submarine mapping project using SimradEM300 multibeam data.



4 of 34

Table 2. Microprobe Analyses of Submarine Kiekie Basalt Glassesa

Sample Numberb SiO2 TiO2 Al2O3 FeO* MnO MgO CaO Na2O K2O P2O5 S Cl Total

T317-R01 6 47.77 1.57 16.08 11.29 0.19 6.89 11.83 3.13 0.36 0.29 0.139 0.026 99.57T317-R01 R 9 47.85 1.56 15.96 11.74 0.18 6.78 11.86 3.07 0.36 0.30 0.132 0.027 99.83T317-R02 5 47.76 1.48 16.42 11.46 0.18 7.26 11.58 2.98 0.35 0.32 0.124 0.021 99.93T317-R03 5 47.73 1.55 15.99 11.46 0.19 6.92 11.84 3.15 0.36 0.29 0.126 0.028 99.65T317-R03 R 6 48.05 1.55 15.90 11.65 0.18 6.81 11.62 3.09 0.36 0.29 0.132 0.027 99.67T317-R04A 6 46.68 1.42 16.39 12.18 0.17 7.80 11.55 2.83 0.32 0.16 0.064 0.032 99.60T317-R08 6 47.81 1.60 16.10 11.44 0.21 6.87 11.84 3.20 0.37 0.30 0.132 0.028 99.88T317-R10 5 45.73 1.96 16.06 12.02 0.19 6.64 12.65 3.20 0.60 0.33 0.172 0.034 99.60T317-R12 13 47.03 1.40 16.61 11.97 0.16 7.86 11.64 2.93 0.31 0.17 0.076 0.021 100.18T317-R13 6 47.68 1.48 16.52 11.29 0.17 7.21 11.80 3.03 0.34 0.28 0.127 0.019 99.96T317-R14 6 46.62 1.47 16.82 11.81 0.17 7.48 11.84 2.98 0.36 0.20 0.108 0.018 99.88T317-R15 6 47.78 1.47 16.42 11.08 0.19 7.24 11.77 3.02 0.34 0.29 0.126 0.021 99.75T317-R15 R 6 48.00 1.48 16.32 11.34 0.17 7.19 11.65 2.97 0.34 0.28 0.127 0.021 99.89T317-R16 6 47.58 1.48 16.37 11.31 0.18 7.24 11.82 3.02 0.35 0.29 0.127 0.022 99.78T317-R18 6 44.95 2.24 15.98 12.04 0.20 6.28 12.99 3.40 0.74 0.40 0.186 0.044 99.45T317-R18 R 6 45.28 2.26 16.12 12.27 0.19 6.31 12.90 3.34 0.73 0.42 0.186 0.042 100.04T318-R01 5 46.21 1.40 16.85 11.76 0.18 7.63 12.19 2.97 0.33 0.15 0.109 0.019 99.80T318-R02 6 46.24 1.32 16.94 11.78 0.20 7.53 12.10 2.95 0.33 0.16 0.107 0.021 99.66T318-R03 6 46.44 1.30 16.99 11.63 0.20 7.39 12.24 2.99 0.33 0.17 0.109 0.022 99.80T318-R04 6 46.26 1.32 16.92 11.73 0.18 7.47 12.18 3.01 0.32 0.16 0.107 0.020 99.68T318-R06 6 46.38 1.33 16.99 11.68 0.19 7.42 12.24 2.98 0.33 0.18 0.111 0.021 99.86T318-R08 6 46.28 1.34 16.93 11.58 0.18 7.48 12.22 2.98 0.33 0.18 0.107 0.022 99.64T318-R09 6 46.25 1.38 16.86 11.84 0.19 7.32 12.22 3.00 0.33 0.16 0.103 0.018 99.65T318-R09 R 10 46.61 1.31 17.05 11.80 0.17 7.42 12.25 2.91 0.32 0.16 0.103 0.020 100.13T318-R10 6 46.24 1.35 16.94 11.73 0.20 7.29 12.23 3.00 0.33 0.17 0.107 0.024 99.63T318-R11 6 46.26 1.31 16.91 11.79 0.21 7.53 12.14 2.97 0.33 0.16 0.106 0.022 99.73T318-R12 6 46.16 1.60 16.60 12.05 0.20 7.46 11.78 3.01 0.39 0.20 0.101 0.019 99.57T318-R13 6 46.39 1.54 16.73 12.05 0.19 7.66 11.79 2.95 0.37 0.19 0.105 0.021 99.99T318-R14 6 46.44 1.50 16.63 12.07 0.19 7.58 11.79 2.96 0.37 0.19 0.106 0.023 99.85T318-R15 5 46.38 1.51 16.68 12.04 0.18 7.56 11.73 2.95 0.36 0.18 0.106 0.018 99.70T318-R16 6 46.39 1.53 16.76 11.88 0.19 7.55 11.79 2.98 0.36 0.19 0.106 0.020 99.75T318-R17 6 46.53 1.56 16.59 12.13 0.20 7.13 11.80 3.03 0.36 0.18 0.106 0.023 99.63T322-R1 6 46.29 1.50 16.50 11.58 0.20 6.97 12.49 3.24 0.48 0.29 0.135 0.024 99.70T322-R2 5 46.67 1.48 16.69 11.72 0.18 6.97 12.55 3.14 0.47 0.29 0.130 0.026 100.31T322-R3 9 47.27 1.44 16.25 11.88 0.19 7.64 11.69 2.72 0.32 0.18 0.071 0.019 99.66T322-R4 6 47.11 1.40 16.73 11.44 0.21 7.35 12.12 3.04 0.37 0.19 0.112 0.019 100.10T322-R5 3 47.13 1.44 16.82 11.46 0.16 7.26 12.21 3.02 0.38 0.19 0.124 0.022 100.22T322-R7 11 47.03 1.40 16.29 11.84 0.16 7.84 11.62 2.88 0.31 0.16 0.080 0.024 99.74T322-R20 10 46.95 1.34 16.58 11.38 0.18 7.92 11.85 2.90 0.32 0.16 0.080 0.018 99.66T322-R30 6 47.30 1.41 16.77 11.68 0.16 7.38 11.60 3.02 0.33 0.17 0.100 0.017 99.92T322-R30 R 5 47.17 1.35 16.63 11.31 0.18 7.40 11.61 2.99 0.32 0.17 0.103 0.015 99.27T322-R31 6 47.31 1.41 16.71 11.83 0.17 7.40 11.57 3.04 0.32 0.17 0.103 0.013 100.06T322-R31 R 5 47.39 1.37 16.83 11.74 0.18 7.49 11.65 3.07 0.31 0.17 0.106 0.017 100.32T322-R32 5 47.30 1.42 16.77 11.94 0.18 7.04 11.72 3.09 0.33 0.18 0.106 0.013 100.09T322-R32 R 6 47.57 1.45 16.66 11.92 0.18 6.97 11.74 2.95 0.33 0.18 0.105 0.017 100.08T322-R33 6 47.29 1.42 16.85 11.78 0.18 7.25 11.62 3.05 0.32 0.18 0.100 0.021 100.07T322-R33 R 5 47.28 1.37 16.80 11.93 0.19 7.34 11.70 3.03 0.31 0.17 0.098 0.017 100.23T323-R1 6 47.14 1.54 16.68 11.16 0.18 6.45 11.81 3.11 0.37 0.27 0.115 0.033 98.87T323-R2 6 47.27 1.58 16.57 11.30 0.16 6.17 11.90 3.17 0.39 0.29 0.124 0.034 98.96T323-R3 6 47.20 1.61 16.30 11.42 0.18 6.05 11.90 3.21 0.40 0.31 0.128 0.038 98.75T323-R4 6 47.37 1.51 16.85 11.64 0.19 6.53 11.79 3.15 0.38 0.28 0.112 0.035 99.85T323-R5 6 47.20 1.52 16.69 11.21 0.18 6.64 11.79 3.12 0.37 0.29 0.115 0.031 99.14T323-R6 6 47.60 1.55 16.86 11.32 0.20 6.58 11.96 3.03 0.37 0.28 0.115 0.035 99.91T323-R8 6 46.65 1.62 16.77 11.54 0.18 6.67 11.97 3.12 0.43 0.28 0.125 0.031 99.39T323-R9 6 47.22 1.51 16.77 11.16 0.19 6.47 11.83 3.14 0.38 0.27 0.108 0.038 99.06T323-R10 6 46.92 1.57 16.69 11.53 0.19 6.97 11.94 3.13 0.44 0.28 0.127 0.034 99.83T323-R11 6 46.61 1.55 16.65 11.57 0.19 6.90 11.94 3.13 0.43 0.28 0.133 0.030 99.42T323-R12 6 47.18 1.55 16.69 11.23 0.18 6.47 11.78 3.12 0.36 0.27 0.111 0.036 98.97T323-R14 6 47.72 1.56 16.62 11.58 0.17 6.60 11.95 3.15 0.38 0.27 0.114 0.035 100.13T323-R15 5 46.72 1.50 16.43 11.41 0.19 6.67 11.82 3.11 0.38 0.28 0.125 0.033 98.66T323-R16 6 47.08 1.56 16.40 11.39 0.18 6.66 11.76 3.15 0.37 0.28 0.122 0.031 98.97T323-R17 5 47.47 1.50 16.49 11.28 0.20 6.62 11.66 3.07 0.37 0.28 0.125 0.033 99.09



5 of 34

infrared spectroscopy (Table 5). Glass chips weredoubly polished to a thickness between about 50and 200 mm. Transmission infrared spectra in the4000–1200 cm�1 (2.5 to 8.3 mm) range werecollected using an infrared microscope attachmentto a Bruker IFS-66 FTIR spectrometer at theUniversity of Miami according to the procedureof Dixon and Clague [2001]. Glass density wascalculated for each sample using the Gladstone-Dale rule and the Church-Johnson equation asdescribed by Silver et al. [1990]. We used a molarabsorptivity of 330 ± 20 L/mole/cm for carbondissolved as carbonate in alkali silicate glassescalculated according to Dixon and Pan [1995].Precision is the about ±2% for total water and±10% for molecular water and carbonate. Theaccuracy of the total water analyses is about±10%. The accuracy of the CO2 and molecularwater analyses are estimated to be about ±20%.

[11] Twelve subaerial Kiekie Basalt samples andeight submarine Kiekie samples, all analyzed formajor and trace elements, were analyzed for Sr,Nd, and Pb isotopic ratios (Table 6) at CarletonUniversity, Ontario using techniques outlined byCousens [1996]. Prior to dissolution, splits for Pband Nd were acid washed in warm 1.5 N HCL for12 hours, then rinsed three times with ultrapureH2O. Splits for Sr were acid-washed in hot 6 NHCl for 4 days, then rinsed three times withultrapure H2O, followed by dissolution in HF/HNO3. Two samples (T317-R18 and T323-R21)were run for Pb following an additional wash inhydroxylamine to eliminate traces of Mn oxides.All Pb mass spectrometer runs were corrected forfractionation using NIST SRM981. The ratiosmeasured for SRM981 are 206Pb/204Pb = 16.890± 0.020, 207Pb/204Pb = 15.429 ± 0.026, and208Pb/204Pb = 36.498 ± 0.048 (2s), based onmultiple runs of standards and duplicates. Thefractionation correction, based on the values ofTodt et al. [1984], is +0.13%/amu. Sr isotopicratios are normalized to 86Sr/88Sr = 0.11940 tocorrect for fractionation. Two Sr standards are runat Carleton, NIST SRM987 (87Sr/86Sr = 0.710251 ±

18) and the Eimer and Amend (E&A) SrCO3

(87Sr/86Sr = 0.708032 ± 24). Nd isotope ratios arenormalized to 146Nd/144Nd = 0.72190 and 54 runs ofthe La Jolla standard average 143Nd/144Nd =0.511876 ± 18.

4. Results

4.1. Major Elements

[12] Major element compositions (Tables 2–4) areplotted on an alkali-silica diagram in Figure 4. Ingeneral, rejuvenated-stage lavas are characterizedby increasing alkalies with decreasing SiO2 attrib-uted to high concentration of incompatible elementsat low extents of partial melting. The scatter in thedata at low SiO2 is caused by either loss of alkaliesduring subaerial weathering or a residual K2O-bearing phase during melting. Kiekie lavas are onlymildly alkalic with higher SiO2 contents than mostother rejuvenated-stage lavas. Also, the subaerialKiekie lavas have distinctly lower alkali contents ata given silica content than the submarine lavas, mostlikely caused by weathering.

4.2. Trace Elements

[13] Trace element compositions are listed inTables 3a, 3b, and 4. Figure 5 shows La plottedagainst MgO, a standard indicator of differentia-tion. Rejuvenated-stage lavas exhibit much greatercompositional variability than shield lavas at agiven MgO. Shield lavas with MgO > 6.5 wt%are characterized by steadily increasing concentra-tions of trace elements with decreasing MgO,interpreted to be the result of olivine fractionation.To remove the effects of olivine fractionation,olivine is typically added until the melt is inequilibrium with an assumed mantle olivine com-position (e.g., Fo90) or has reached an assumedprimitive melt MgO. Olivine addition programswork well for tholeiitic compositions; however,uncertainties in the correction increase significantlyas melt compositions become more alkalic. Oneimportant variable is oxygen fugacity (melt Fe3+/Fe2+);

Table 2. (continued)

Sample Numberb SiO2 TiO2 Al2O3 FeO* MnO MgO CaO Na2O K2O P2O5 S Cl Total

T323-R18 5 47.78 1.54 16.56 11.27 0.20 6.44 11.82 3.11 0.37 0.27 0.125 0.030 99.52T323-R19 5 47.72 1.52 16.57 11.37 0.17 6.64 11.73 3.05 0.37 0.28 0.123 0.031 99.57T323-R20 5 47.81 1.55 16.62 11.50 0.17 6.74 11.83 3.05 0.37 0.28 0.125 0.033 100.06T323-R21 5 47.80 1.70 15.69 12.03 0.19 5.95 11.81 3.24 0.42 0.31 0.140 0.037 99.31

aR indicates repeat analysis done during different microprobe session.

bNumber indicates number of individual point analyses included in average.



6 of 34

Table

3a.

MajorandTrace

ElementAnalysesofSubmarineKiekie

Basalt

Sam

ple

T317-

R2

T317-

R10

T317-

R14

T317-

R18

T318-

R2

T318-

R3

T318-

R11

T318-

R13

T318-

R17

T318-

R19

T322-

R1

T322-

R4

T322-

R6

T322-

R31

T322-

R32

T323-

R3

T323-

R9

T323-

R11

T323-

R21

SiO

247.69

46.46

47.17

45.46

46.23

46.33

46.25

46.08

45.81

43.12

47.04

46.95

46.99

47.04

46.75

47.42

47.6

47.08

48.13

TiO

21.43

1.98

1.33

2.16

1.19

1.18

1.17

1.39

1.31

2.23

1.32

1.22

1.43

1.29

1.25

1.31

1.32

1.42

1.6

Al2O3

15.75

15.95

15.05

15.91

15.35

15.03

15.31

15.24

15.13

12.22

16.1

15.08

17.28

15.56

14.95

14.75

15.08

15.24

16.31

FeO

*11.55

11.47

11.61

12.16

11.55

11.17

12.13

11.9

12.28

12.72

10.64

11.03

9.94

11.33

11.99

11.43

11.65

11.83

11.61

MnO

0.17

0.17

0.18

0.18

0.2

0.18

0.2

0.18

0.19

0.19

0.19

0.19

0.15

0.18

0.19

0.18

0.18

0.18

0.19

MgO

8.27

6.76

10.47

7.02

11.42

12.02

11.25

10.87

11.09

13.27

9.71

11.65

7.52

10.63

11.2

11.82

10.93

10.64

6.55

CaO

11.45

12.74

10.81

12.83

11.03

10.95

11.02

10.86

10.66

12.03

11.84

10.91

13.07

10.84

10.45

10.44

10.63

10.86

11.9

Na 2O

2.78

2.79

2.6

2.75

2.43

2.31

2.36

2.5

2.52

2.65

2.73

2.52

2.86

2.62

2.51

2.48

2.54

2.62

2.89

K2O

0.44

0.8

0.34

0.77

0.32

0.44

0.38

0.43

0.38

1.02

0.54

0.38

0.42

0.34

0.34

0.59

0.47

0.53

0.62

P2O5

0.31

0.32

0.26

0.38

0.16

0.16

0.15

0.18

0.17

0.46

0.27

0.18

0.3

0.17

0.17

0.25

0.25

0.27

0.3

Total

99.84

99.44

99.82

99.63

99.88

99.77

100.21

99.63

99.55

99.91

100.37

100.12

99.96

100

99.8

100.66

100.65

100.67

100.09

XRFTrace

ElementAnalyses

Ni

172

95

224

77

285

334

293

272

269

327

171

255

92

270

281

317

287

261

64

Cr

373

231

459

199

359

389

352

351

350

531

275

473

164

366

372

521

492

431

147

V259

306

224

305

250

235

238

241

235

337

252

233

277

237

226

232

238

257

281

Ga

16

18

17

17

17

17

18

15

18

17

16

15

21

16

16

14

16

17

19

Cu

87

85

84

61

100

98

103

92

97

79

90

84

97

99

91

82

86

90

102

Zn

88

86

87

86

79

83

78

82

80

92

79

81

78

84

78

86

88

86

88

ICP-M

STrace

ElementAnalyses

La

12.38

17.96

10.81

22.26

7.60

8.15

7.49

9.16

8.16

22.28

12.68

7.99

16.15

7.69

7.53

10.17

10.52

11.69

13.00

Ce

25.07

35.54

22.33

43.65

15.59

16.45

15.30

18.99

16.99

44.85

25.06

16.46

31.42

16.26

15.85

21.21

21.68

24.19

27.07

Pr

3.37

4.65

3.03

5.64

2.11

2.19

2.07

2.57

2.33

5.70

3.23

2.22

3.96

2.25

2.20

2.87

2.96

3.27

3.67

Nd

14.98

20.06

13.43

24.13

9.52

9.74

9.32

11.48

10.50

23.96

13.75

9.85

16.90

10.17

9.90

12.69

13.13

14.43

16.18

Sm

3.70

4.79

3.39

5.50

2.62

2.66

2.54

2.97

2.78

5.42

3.37

2.69

3.99

2.78

2.69

3.21

3.33

3.54

4.10

Eu

1.28

1.60

1.19

1.86

0.96

0.94

0.92

1.07

1.01

1.73

1.16

0.96

1.36

1.00

0.97

1.13

1.17

1.22

1.41

Gd

3.91

4.91

3.69

5.33

2.96

3.00

2.91

3.25

3.11

4.95

3.48

3.06

4.08

3.13

3.04

3.45

3.54

3.68

4.34

Tb

0.60

0.73

0.57

0.77

0.50

0.49

0.48

0.52

0.50

0.70

0.54

0.49

0.63

0.50

0.49

0.54

0.56

0.58

0.69

Dy

3.44

4.23

3.36

4.41

3.01

3.01

2.96

3.13

3.06

3.70

3.18

2.96

3.72

3.08

2.98

3.19

3.32

3.42

3.99

Ho

0.68

0.81

0.66

0.81

0.63

0.61

0.61

0.64

0.62

0.67

0.64

0.59

0.74

0.62

0.60

0.63

0.66

0.67

0.80

Er

1.85

2.16

1.79

2.16

1.75

1.70

1.68

1.75

1.74

1.70

1.77

1.64

1.96

1.69

1.64

1.68

1.77

1.83

2.13

Tm

0.26

0.31

0.25

0.30

0.26

0.25

0.25

0.26

0.25

0.22

0.25

0.24

0.29

0.25

0.24

0.24

0.25

0.26

0.31

Yb

1.61

1.91

1.57

1.85

1.65

1.63

1.61

1.63

1.59

1.30

1.63

1.52

1.89

1.56

1.52

1.51

1.56

1.62

1.92

Lu

0.26

0.30

0.24

0.29

0.27

0.26

0.26

0.26

0.26

0.20

0.25

0.24

0.29

0.25

0.24

0.23

0.24

0.25

0.30

Ba

240

313

208

395

127

121

124

165

152

317

217

140

257

145

142

195

197

236

244

Th

0.80

1.35

0.68

1.79

0.57

0.61

0.55

0.60

0.52

1.71

0.95

0.58

1.22

0.49

0.47

0.59

0.59

0.70

0.74

Nb

10.46

19.84

9.79

24.51

8.85

8.74

8.70

8.85

8.14

28.39

14.35

10.01

16.50

7.79

7.59

8.62

8.83

10.63

10.95

Y20.80

24.28

19.68

25.09

18.20

17.77

17.89

18.42

18.16

19.65

18.82

17.61

22.20

17.98

17.49

18.28

19.09

19.44

23.45

Hf

1.39

2.07

1.36

2.36

1.27

1.27

1.27

1.33

1.25

2.82

1.44

1.34

1.62

1.22

1.18

1.27

1.31

1.36

1.59

Ta

0.56

1.12

0.53

1.40

0.52

0.52

0.52

0.53

0.47

1.77

0.78

0.56

0.88

0.46

0.44

0.48

0.48

0.59

0.59

U0.24

0.32

0.16

0.38

0.14

0.24

0.13

0.27

0.13

0.49

0.23

0.15

0.37

0.21

0.12

0.19

0.16

0.19

0.19



7 of 34

strongly alkalic compositions tend to be moreoxidized than tholeiitic ones [Dixon et al., 1997].For example, North Arch basalt with 7 wt% MgOat NNO-2 requires addition of 31% olivine, result-ing in a primitive melt MgO content of 16.3 wt%to be in equilibrium with Fo90. Under more oxi-dizing conditions (NNO + 1), the amount ofolivine added is reduced to 16%, resulting in aprimitive melt MgO content of 12.5 wt%. In theabsence of Fe3+/Fe2+ measurements and given theuncertainties in conventional olivine addition pro-grams, we corrected our data using a simple linearcorrection to MgO = 14 wt%. The solid line onFigure 5 shows this olivine correction, with slopedefined by �2.8%/(1 wt% MgO), consistent withthe slope of shield basalts and the results ofolivine-addition calculations. Many rejuvenated-stage lavas require little correction. For NorthArch, Kiekie submarine and South Arch lavas,correction from 7 to 14 wt% results in a reductionof only �20% in all trace elements.

[14] Figures 6–12 display correlations among thetrace element data. Trace element variations dis-cussed below will be shown as a function of Th,the most incompatible element in this setting.Figures 6a, 7a, 8a, 9a, 10a, 11a, and 12a (upperleft-hand side) show the entire concentration rangefor all rejuvenated-stage lavas, while Figures 6b,7b, 8b, 9b, 10b, 11b, and 12b (lower left-hand side)focuses on the Kiekie lavas (note scale change).Figures 6c–6d, 7c–7d, 8c–8d, 9c–9d, 10c–10d,11c–11d, and 12c–12d (right-hand side) show theMg14-corrected data for samples having MgO >6.5 wt% along with model curves as described insection 5.

[15] Figures 6a and 6b show Ba plotted against Thfor Hawaiian shield and rejuvenated-stage lavas.Both Ba and Th are highly incompatible; thereforea positive correlation is expected for samplesrelated simply by magmatic processes, such asvariable extents of melting of a homogeneoussource region, variable enrichment or depletionof the source caused by addition or subtraction ofmelts, or extents of crystallization of a singleparent magma. Such a correlation is observed forthe tholeiitic shield lavas. Most rejuvenated-stagelavas have higher concentrations of both Ba and Ththan the shield lavas. The nephelinites and melili-tites from the subaerial Honolulu Volcanics havethe highest Ba and Th, but a slightly lower Ba/Thratio. Kiekie lavas have Th concentrations <2 ppm,lower than other rejuvenated-stage lavas (1 to12 ppm), but have Ba concentrations significantly

Table

3a.(continued)

Sam

ple

T317-

R2

T317-

R10

T317-

R14

T317-

R18

T318-

R2

T318-

R3

T318-

R11

T318-

R13

T318-

R17

T318-

R19

T322-

R1

T322-

R4

T322-

R6

T322-

R31

T322-

R32

T323-

R3

T323-

R9

T323-

R11

T323-

R21

Pb

0.77

1.14

0.66

1.62

0.54

0.58

0.55

0.62

0.55

1.68

0.78

0.54

0.98

0.51

0.48

0.58

0.59

0.66

0.74

Rb

8.35

17.10

5.40

16.90

5.95

7.60

9.70

6.80

6.10

21.00

9.50

6.00

8.60

5.20

5.30

6.80

6.80

9.90

6.70

Cs

0.23

0.56

0.08

0.45

0.09

0.19

0.61

0.12

0.11

0.24

0.17

0.09

0.09

0.10

0.07

0.17

0.26

0.42

0.09

Sr

539

566

473

709

270

275

276

346

319

589

416

299

518

302

299

381

397

403

448

Sc

31.8

37.4

30.6

35.1

32.8

29.0

31.0

29.9

28.9

29.4

29.7

30.3

33.8

29.3

27.9

26.4

27.4

29.3

33.2

Zr

53

81

51

96

46

44

45

46

44

110

57

51

65

43

42

46

47

50

58



8 of 34

in excess of the shield lavas (Figure 6b). Lavascollected furthest from the island (T323 lavas) havethe greatest enrichments in Ba. Several other sam-ples, including South Arch and three submarineHonolulu lavas (T273 lavas), are enriched in Basimilar to the Kiekie lavas. These lavas will bereferred to as high-Ba lavas.

[16] High-Ba lavas are also enriched in Sr. Reju-venated-stage lavas form a linear array extendingto higher Sr and Th concentrations than shieldlavas, but with a lower Sr/Th (Figure 7a). Kiekieand other high-Ba lavas have Sr and Th concen-trations that overlap with the shield lavas, signifi-cantly enriched in Sr relative to an extrapolationthrough the trend formed by the other rejuvenated-stage lavas (Figure 7b). Clague et al. [2006]showed that similar Sr enrichments in the subma-rine Honolulu Series lavas can not be explained byseawater addition (<5% of Sr added by seawater).

[17] In contrast to Ba and Sr, high-field strengthelement concentrations in high-Ba samples showno enrichment relative to other rejuvenated-stage

lavas. Kiekie lavas have low Nb and Hf concen-trations consistent with their low Th concentrationsand the trend formed by the other rejuvenated-stagelavas (Figures 8a and 8b and Figures 9a and 9b).South Arch and three high-Ba Honolulu Volcanicslavas, however, have Nb and Hf concentrationsslightly elevated above Kiekie lavas.

[18] La (Figure 10) and Lu (Figure 11) illustrate therange of behavior of the rare-earth elements (REE).Relative to the trends formed by representativerejuvenated-stage lavas, high-Ba lavas have higherconcentrations of both light rare-earth (LREE) andheavy rare-earth (HREE) elements.

[19] Figure 12 shows Nb/La plotted against Ba/Th.It has long been recognized that Hawaiian lavas arecharacterized by higher Ba/Th than other oceanicisland basalts (OIB) [e.g., Hofmann and Jochum,1996; Yang et al., 2003]. Hawaiian shield lavascluster around a Ba/Th value of 132 ± 20. Incontrast, Easter Seamount Chain lavas [Fontignieand Schilling, 1991; Kingsley, 2002; Kingsley etal., 2007] whose compositions are not shown butare representative of other Pacific plumes (e.g.,Galapagos and Pukapuka), have significantly lowerBa/Th ranging from �55 for the more depletedsamples to �70 for the more enriched samples.Most Hawaiian rejuvenated-stage lavas form afield with Ba/Th (153 ± 28) and Nb/La (1.43 ±0.15), higher than the shield lavas. In contrast,Kiekie lavas form a distinct trend toward higherBa/Th and lower Nb/La. This trend is consistentwith the concentration data showing that Kiekielavas are enriched in Ba and La, but not in Nb.High Ba/Th in the T323 submarine Kiekie lavas(up to 353) are the highest values ever measured inHawaiian lavas. Though Kiekie lavas define themost extreme values, the signature is not limited toKiekie. Honolulu Volcanics, Kaula nephelinites,and South Arch lavas all have higher Ba/Th andlower Nb/La than the representative Hawaiianrejuvenated-stage lavas.

4.3. Water, Carbon Dioxide, and Chlorine

[20] Dissolved water and carbon dioxide concen-trations in quenched glassy rinds of the submarinelavas are listed in Table 5. Depth of vapor satura-tion and composition of the equilibrium vaporphase based on dissolved water and carbon dioxideconcentrations are calculated according to Dixon[1997].

[21] Figure 13 shows the depth of vapor saturationplotted against the depth of sample collection.Most rejuvenated-stage lavas, including Kiekie

Table 3b. ICP-MS Trace Element Analyses ofSubmarine Glasses

SampleT318-R8

T318-R9

T323-R8

T323-R10

T323-R12

T323-R15

T323-R16

La 8.54 7.45 11.86 12.16 10.84 11.09 11.65Ce 17.36 15.38 24.68 25.33 22.62 22.98 24.33Pr 2.28 2.09 3.33 3.38 3.06 3.08 3.29Nd 10.32 9.40 14.63 14.81 13.61 13.57 14.59Sm 2.75 2.57 3.59 3.66 3.41 3.43 3.68Eu 0.99 0.92 1.25 1.26 1.20 1.23 1.29Gd 3.14 2.97 3.81 3.84 3.62 3.61 3.96Tb 0.52 0.49 0.59 0.59 0.57 0.58 0.62Dy 3.18 3.05 3.44 3.47 3.36 3.39 3.61Ho 0.65 0.61 0.67 0.69 0.67 0.68 0.71Er 1.80 1.74 1.84 1.85 1.77 1.79 1.89Tm 0.27 0.26 0.26 0.27 0.26 0.26 0.27Yb 1.73 1.67 1.66 1.68 1.64 1.63 1.68Lu 0.27 0.26 0.26 0.26 0.25 0.25 0.26Ba 135 127 237 244 206 206 218Th 0.63 0.57 0.68 0.69 0.61 0.63 0.66Nb 9.52 8.93 10.99 11.15 9.31 9.25 9.59Y 19.33 18.40 20.33 20.64 20.12 19.97 20.52Hf 1.34 1.29 1.43 1.43 1.36 1.37 1.41Ta 0.50 0.48 0.56 0.56 0.47 0.47 0.49U 0.17 0.15 0.18 0.18 0.16 0.16 0.16Pb 0.63 0.62 0.72 0.73 0.64 0.69 0.70Rb 7.0 6.9 7.2 7.4 5.9 5.9 5.9Cs 0.09 0.09 0.09 0.09 0.06 0.06 0.08Sr 297 266 420 427 416 417 428Sc 32.9 33.6 30.5 30.9 29.5 30.5 26.5Zr 49 46 53 53 50 49 51



9 of 34

Table

4(Sample).Wet

Chem

ical

MajorElementandTrace

ElementAnalysesofSubaerial

Kiekie

Basalt[ThefullTable

4isavailable

intheHTMLversionof

thisarticleat

http://www.g-cubed.org]

Sam

ple

69Nii-1

69Nii-3

69Nii-4

69Nii-5

69Nii-6

69Nii-7

69Nii-8

69Nii-9

70Nii-1

70Nii-4

70Nii-5

70Nii-6

70Nii-7

70Nii-8

70Nii-9

70Nii-10

70Nii-11

70Nii-12

SiO

245.64

45.50

46.26

45.35

45.27

46.18

46.97

45.42

44.73

46.74

46.25

44.72

44.99

45.49

45.21

45.79

45.31

45.19

TiO

21.16

1.07

1.10

1.35

1.15

1.18

1.38

1.12

1.38

1.48

1.44

1.17

1.25

1.36

1.32

1.49

1.28

1.28

Al 2O3

15.23

15.22

15.40

15.34

14.92

15.60

16.78

15.21

14.82

15.00

15.10

15.04

15.53

15.24

15.32

16.38

15.25

15.33

Fe 2O3

4.10

5.28

4.80

3.37

5.05

2.03

2.35

1.74

4.48

1.95

9.13

2.26

2.72

2.15

2.73

2.34

3.40

3.97

FeO

7.98

6.84

6.72

8.37

6.75

9.36

8.62

9.72

7.74

9.72

3.33

9.18

8.96

9.63

8.78

9.25

8.34

7.76

MnO

0.19

0.19

0.18

0.19

0.19

0.18

0.18

0.18

0.18

0.17

0.18

0.18

0.18

0.18

0.17

0.18

0.18

0.18

MgO

11.34

10.65

10.00

10.18

12.20

10.75

7.44

11.81

10.22

10.03

10.13

11.29

11.42

11.33

10.71

9.99

11.07

10.16

CaO

10.69

11.05

11.23

11.08

10.86

11.31

12.49

10.78

11.08

11.02

10.72

11.34

11.04

10.83

11.09

11.15

10.93

11.20

Na 2O

2.46

2.36

2.40

2.84

2.34

2.56

2.49

2.39

2.53

2.68

2.53

2.61

2.30

2.60

2.76

2.80

2.49

2.80

K2O

0.29

0.22

0.21

0.37

0.31

0.28

0.31

0.25

0.29

0.29

0.28

0.34

0.31

0.33

0.39

0.39

0.34

0.39

P2O5

0.19

0.12

0.13

0.30

0.18

0.17

0.19

0.18

0.18

0.20

0.21

0.25

0.26

0.25

0.27

0.27

0.24

0.29

H2O+

0.06

0.30

0.39

0.31

0.39

0.10

0.40

0.31

0.72

0.06

0.22

0.21

0.57

0.19

0.21

0.25

0.43

0.29

H2O�

0.21

0.53

0.47

0.35

0.28

0.13

0.23

0.30

0.91

0.26

0.28

0.37

0.38

0.22

0.39

0.36

0.39

0.48

CO2

0.00

0.01

0.01

0.01

0.01

0.00

0.00

0.11

0.04

0.00

0.00

0.35

0.19

0.02

0.01

0.10

0.08

0.00

S0.06

0.14

0.17

0.13

0.02

0.03

0.03

0.12

0.11

0.08

0.02

0.18

0.03

0.03

0.12

0.05

0.04

0.20

Total

99.71

99.60

99.59

99.63

100.04

99.96

99.93

99.75

99.53

99.79

99.94

99.61

100.22

99.97

99.58

100.90

99.89

99.64

XRFTrace

ElementAnalyses

Ni

274

275

265

233

340

229

116

291

268

190

221

250

267

264

296

200

268

234

Cr

416

474

482

381

521

426

308

452

442

423

468

409

443

408

369

365

420

388

V215

194

191

216

215

205

219

201

281

228

236

205

211

229

249

247

200

216

Ga

17

18

18

18

17

18

20

17

17

18

18

17

18

17

16

18

17

17

Cu

95

100

120

90

70

85

100

100

100

85

80

70

37

75

70

75

70

75

Zn

87

77

82

79

83

75

74

88

102

92

102

78

78

81

107

106

79

83

Co

63

66

60

58

69

60

48

62

60

55

59

58

61

60

59

55

60

59

ICP-M

STrace

ElementAnalyses

La

8.91

8.04

7.78

16.84

13.22

9.19

10.74

8.34

10.72

9.53

8.99

13.67

13.84

15.97

15.67

16.52

14.21

16.61

Ce

18.35

15.22

14.68

32.87

22.15

17.69

21.51

17.30

19.58

20.21

19.58

27.01

27.49

28.15

30.70

32.73

28.21

32.48

Pr

2.47

2.16

2.16

4.19

3.02

2.44

2.88

2.32

2.77

2.84

2.70

3.46

3.55

3.86

3.90

4.25

3.69

4.19

Nd

10.84

9.92

9.94

17.60

13.26

11.03

12.90

10.26

12.57

13.10

12.44

14.93

15.18

16.65

16.59

17.83

15.62

17.67

Sm

2.84

2.69

2.77

3.99

3.22

2.91

3.47

2.71

3.34

3.57

3.41

3.51

3.53

3.86

3.76

4.24

3.61

4.06

Eu

1.03

0.98

1.03

1.36

1.14

1.08

1.27

1.00

1.19

1.29

1.23

1.20

1.23

1.35

1.30

1.44

1.23

1.36

Gd

3.17

3.19

3.33

3.98

3.64

3.30

3.95

3.00

3.70

3.99

3.75

3.58

3.66

4.04

3.75

4.17

3.63

3.90

Tb

0.51

0.51

0.53

0.59

0.56

0.52

0.62

0.47

0.58

0.62

0.59

0.54

0.55

0.61

0.56

0.63

0.54

0.58

Dy

3.02

3.11

3.24

3.45

3.29

3.13

3.69

2.87

3.49

3.74

3.51

3.11

3.15

3.66

3.27

3.63

3.20

3.37

Ho

0.60

0.65

0.65

0.67

0.67

0.63

0.74

0.57

0.70

0.73

0.68

0.62

0.63

0.74

0.63

0.70

0.63

0.66

Er

1.64

1.83

1.75

1.77

1.81

1.68

1.93

1.53

1.89

1.93

1.85

1.66

1.69

2.02

1.66

1.87

1.69

1.78

Tm

0.24

0.28

0.25

0.26

0.26

0.24

0.28

0.22

0.28

0.27

0.26

0.24

0.24

0.29

0.24

0.27

0.24

0.26

Yb

1.55

1.69

1.57

1.62

1.61

1.54

1.77

1.45

1.72

1.72

1.65

1.52

1.53

1.78

1.51

1.68

1.54

1.57

Lu

0.25

0.27

0.24

0.25

0.26

0.24

0.27

0.22

0.28

0.27

0.25

0.23

0.25

0.28

0.24

0.26

0.24

0.25

Ba

147

116

109

264

229

154

184

141

156

135

132

233

219

219

252

273

227

264

Th

0.60

0.47

0.45

1.13

0.69

0.65

0.77

0.59

0.73

0.62

0.58

0.91

0.95

0.94

1.14

1.12

0.93

1.12



10 of 34

Basalts, lie within uncertainty (±25%) of the 1:1line suggesting that melts were in equilibrium witha mixed CO2-H2O vapor phase during eruption andlittle transport has occurred since eruption. This isimportant because only samples erupted deep onthe seafloor are likely to have retained all or mostof their original water contents. Kiekie sampleT317R14 is slightly undersaturated and may haveerupted at a shallower depth followed by transportby flow to their present collection depth. Twosamples (Kiekie T317-R16A and North Arch23D) are slightly oversaturated for their depthof eruption and may have had insufficient timeto maintain vapor-saturation during eruption.(Table 6).

[22] Water and chlorine concentrations show alarge amount of scatter at a given Th concentration(Figures 14 and 15, respectively). The most strik-ing feature of Figure 14b is the large range of waterconcentrations at a constant Th for Kiekie T323and South Arch lavas. Though both Kiekie T323and South Arch lavas are enriched in Ba, Sr, andLREE, the enrichments in water and trace elementsdo not correlate in any simple way. The trendtoward high water concentrations at nearly constantTh is extremely rare in MORB and OIB basalts.The only other published oceanic island basalt withsimilarly high water concentrations is in meltinclusions within olivine phenocrysts from Loihi,

interpreted to be due to assimilation of seawater[Kent et al., 1999]. Typically, such large variationsof water at relatively constant Th are observed inarc and back-arc environments and interpreted tobe the result of subduction-related fluid fluxing[e.g., Sisson and Layne, 1993; Stolper and Newman,1994;Danyushevsky et al., 1993; Kamenetsky et al.,1997; Newman et al., 2000; Walker et al., 2003].

[23] Some representative rejuvenated-stage lavas,specifically the strongly alkalic North Arch lavas,have water concentrations lower than expected.These low water concentrations have been previ-ously modeled as water loss during degassing[Dixon et al., 1997]. Degassing of water fromsubmarine basalts is usually associated with somecombination of shallow eruption depth, high ve-sicularity, or volatile-rich compositions due to lowextents of melting or high extents of differentiation[e.g., Dixon et al., 1997; Dixon and Clague, 2001;Simons et al., 2002]. None of these factors apply tothe Kiekie basalts.

[24] High-Ba lavas are characterized by high andhighly variable H2O/Ce. High water concentrationsin the Kiekie T323 and South Arch samples result inanomalously high H2O/Ce, up to 620, the highestvalue measured in OIB lavas. Representative sub-

Figure 4. Total alkalies versus silica diagram showingwhole-rock and glass analyses for Kiekie lavas. Kiekielavas are mildly alkalic compared to other Hawaiianrejuvenated-stage lavas. Representative rejuvenated-stage lavas (RSV) include submarine lavas from theKauai-Oahu Channel and the North Arch and subaeriallavas from West Maui (Lahaina Basalt), East Molokai(Kalaupapa Basalt), and Kauai (Koloa Volcanics).Kiekie Submarine T323 samples are from the furthestdive location from the island of Niihau (Figure 3). SeeTable 1 for sources of data.

Figure 5. MgO-variation diagram for La in shield andrejuvenated-stage lavas. Symbols are the same as inFigure 4 with the addition of high-MgO glasses fromKilauea’s Puna Ridge [Clague et al., 1991]. Solid linerepresents a linear approximation of variation producedby olivine-addition with a slope of �2.8% for every1 wt% increase in MgO. Note that magnitude of traceelement variation in rejuvenated-stage lavas is an orderof magnitude greater than the variation produced byolivine fractionation.



11 of 34

marine rejuvenated-stage lavas have H2O/Ce con-sistent with those of Pacific MORB and other OIB(Easter Seamount Chain (210 ± 20) [Simons et al.,2002], Kilauea 41D (196) [Dixon and Clague,2001], and Loihi (173 ± 26) [Dixon and Clague,2001]).

[25] High-Ba lavas, especially Kiekie T323 andSouth Arch, are also enriched in Cl (Figures 15aand 15b) relative to representative rejuvenated-stage lavas, although there is also considerablescatter in the representative rejuvenated-stage lavas(mainly North Arch lavas).

Figure 6. Ba versus Th for Hawaiian shield and rejuvenated-stage lavas. (a) Uncorrected data for lavas with MgO >6.0 wt%. (b) Close-up of low Th concentration data (note scale change) showing that Kiekie lavas are enriched in Baat a given Th concentration relative to other rejuvenated-stage lavas. (c) Ba and Th corrected to MgO = 14 wt% (Ba14,Th14) using the linear correction from Figure 5. Because olivine addition correction was performed only on lavas withMgO > 6.5%, some data on Figures 6a and 6b (in particular representative RSV with 1 < Th < 1.5 ppm), do notappear on Figures 6c and 6d. Solid blue lines represent 1% to 20% batch melting of a metasomatized mantle source(Model 2, 98% depleted mantle +2% of a 0.5% incipient melt of the Hawaiian mantle). Batch melting calculationswere performed using variable mantle mineralogy with 3% residual garnet (crosses), 6% residual garnet (opencircles), and 3% residual garnet along with phlogopite and an oxide phase (open triangles). See Tables 7a, 7b, and 7cfor details. Symbols on the batch melting curves represent 1% to 20% melting at increments of 1% melting. Orangesolid line represents 1% to 20% batch melting of a more depleted source (Model 1, 99.2% depleted mantle + 0.8% ofa 0.5% incipient melt of the Hawaiian mantle). Orange dashed lines represent 1% to 20% batch melting of Model 2mantle to which 0.1% (open square with dot) and 0.2% (open square with cross) carbonatite melt has been added. (d)Close-up of Kiekie data. Because Ba and Th have similar bulk distribution coefficients, high Ba at low Thconcentrations in Kiekie lavas can not be explained by simple melt metasomatism models. Other high-Ba lavasinclude South Arch and T273 Honolulu Volcanics. Addition of up to 0.2% carbonatite to Model 1 mantle source canexplain preferential enrichment of Ba relative to Th.



12 of 34

[26] In summary, Kiekie Basalts have lower incom-patible trace element concentrations than otherrejuvenated-stage lavas. Relative to the trend ex-trapolated through other rejuvenated-stage lavas tolow Th contents, however, Kiekie Basalts areenriched in Ba, Sr, LREE, H2O, and Cl. Noenrichments in Nb, Hf, and Ta are observed.

4.4. Radiogenic Isotopes

[27] In general, the Kiekie Basalts have radiogenicisotopic compositions similar to other rejuvenated-stage lavas in that their compositions are shiftedaway from Hawaiian shield lava compositionstoward a depleted component (Figure 16). Kiekielavas are shifted to higher 87Sr/86Sr and lower206Pb/204Pb than rejuvenated-stage lavas eruptedon other islands and marginal seafloor settings.

5. Modeling

5.1. Melt Metasomatism and MantleSources for Representative Rejuvenated-Stage Lavas

[28] Early modeling of formation of Hawaiianrejuvenated-stage lavas focused on metasomatismof OIB mantle by low degree partial melts from aMORB source [Chen and Frey, 1983, 1985;

Feigenson, 1984; Clague and Dalrymple, 1988;Chen et al., 1990; Maaløe et al., 1992]. Recentradiogenic isotopic studies [Bizimis et al., 2005;Frey et al., 2005; Salters et al., 2006; Fekiacova etal., 2007] show that the depleted component can-not be Pacific lithosphere (either young or aged),and instead is intrinsic to the Hawaiian plume.Other models have shown that trace element andradiogenic isotopic compositions of rejuvenated-stage lavas can be generated from depleted mantlerecently metasomatized by incipient melts (<2%)from the Hawaiian plume [Roden et al., 1984;Reiners and Nelson, 1998; Clague et al., 1990;Dixon et al., 1997; Frey et al., 2000; Yang et al.,2003]. In this metasomatized source, incompatibleelement abundances, as well as Sr and Nd isotopicratios, are dominated by incipient melts.

[29] Our modeling builds on the work of Yang etal. [2003]. We use the depleted MORB sourcecomposition ofWorkman and Hart [2005]. Hawaiianmantle composition is estimated such that the com-positional range of Hawaiian shield lavas is gener-ated by 4–15% batch melting using accepteddistribution coefficients and mantle modal mineral-ogy (Tables 7a, 7b, and 7c). Our Hawaiian mantlecomposition is consistent with estimates of Reinersand Nelson [1998] and Huang et al. [2005] for thesubset of elements evaluated by those authors. We

Figure 7. (a and b) Sr versus Th and (c and d) Sr14 versus Th14, with format following Figure 6. Kiekie and otherhigh-Ba lavas are also enriched in Sr. High Sr concentrations in Kiekie Basalts, South Arch, and Honolulu VolcanicsT273 lavas can be explained by addition of 0.1 to 0.2% carbonatite to the more depleted Model 1 mantlecomposition.



13 of 34

recognize that the Hawaiian mantle is lithologicallyand geochemically heterogeneous, but our estimatedcomposition represents an average value that gen-erates the overall compositional trends in Hawaiianshield tholeiites. We also recognize that batchmelting is an overly simplistic mechanism, butbelieve that it is a valid approximation for aggre-gated fractional melting and is useful for determin-ing bulk F and general trends (see discussion in thework of Plank and Langmuir [1992]).

[30] We generated a range of metasomatized man-tle compositions by adding small amounts ofincipient melts to the depleted MORB source(Tables 7a, 7b, and 7c). Though a larger composi-tional space was examined, we show only theresults for the two most successful mantle compo-sitions: Model 1 has 99.2% depleted mantle +0.8% melt (0.5% melt of Hawaiian plume). Model2 has 98% depleted mantle + 2% melt (0.5% meltof Hawaiian plume). These calculations are meantto illustrate a family of plausible models. As withany such model, there are uncertainties in bulkdistribution coefficients, end-member mantle com-positions, and extent of melting for both lavas and

incipient melts, leading to uncertainties in the finalcalculated product. Nevertheless, it is possible toconstruct a detailed model that is internally consis-tent, as well as consistent with a large body ofpublished data, including experimental data ondistribution coefficients and melting relations.

[31] Representative rejuvenated-stage lavas arebest modeled by 0.8% to 9% batch melting ofModel 2 mantle composition. Figures 6–12 andFigures 14–15 show the results of 1–20% batchmelting of this metasomatized mantle composition(solid blue curves). Batch melting models areshown using a range of mantle mineralogies, whichhighlight the effects of variable garnet in the source(3% and 6%) and the presence of residual phlog-opite and/or oxides such as ilmenite or titanite. Seecaption to Tables 7a, 7b, and 7c for details.Consistent with previous studies, the range incompositions of the representative rejuvenated-stage lavas can be modeled by variable degreesof partial melting of garnet-bearing lherzolite leav-ing residual apatite, phlogopite, and titanite at thesmallest degrees of melting that produce nephelin-ites and nepheline melilitites [e.g., Clague and

Figure 8. (a and b) Nb versus Th and (c and d) Nb14 versus Th14, with format following Figure 6. Nb correlatespositively with Th for most rejuvenated-stage lavas. The presence of garnet does not significantly change the bulkdistribution coefficient of Nb; therefore melting curves of garnet (3%) and garnet (6%) lherzolite mineralogies areindistinguishable. The mantle source of the high-Th Honolulu Volcanics (lowest extent partial melts) most likelycontains an additional accessory phase capable of retaining Nb. Kiekie Basalts contain Nb and Th concentrationslower than other rejuvenated-stage basalts, consistent with larger extents of melting of a more depleted source.Because carbonatites contain low concentrations of high-field strength elements, addition up to 0.2% carbonatite tothe Model 1 mantle source has negligible effect on model melt compositions (Figures 8c and 8d) for the HFSE.



14 of 34

Figure 9. (a and b) Hf versus Th and (c and d) Hf14 versus Th14, with format following Figure 6. Because Hf ismore compatible than Th, addition of incipient melt of Hawaii plume shifts mantle source composition to lower Hf/Th values. Representative rejuvenated-stage lavas are bounded by melts of Model 2 mantle of garnet lherzolite with3% to 6% residual garnet. Low Hf concentrations in Kiekie Basalts require greater extents of melting (4 to 13%) of amore depleted source (Model 1). Addition of carbonatite results in negligible change in the Model 1 melting modelsfor the HFSE.

Figure 10. (a and b) La versus Th and (c and d) La14 versus Th14, with format following Figure 6. La correlatespositively with Th for most rejuvenated-volcanics with a La/Th slightly lower than shield lavas. Kiekie Basalts,South Arch, and T273 Honolulu Volcanics lavas (Figure 10b) have higher La than expected for their low Thconcentrations, consistent with addition of up to 0.2% carbonatite to Model 1 mantle source.



15 of 34

Frey, 1982; Yang et al., 2003; Clague et al., 2006].Concentrations of all incompatible elements (in-cluding Ba, Sr, LREE, HFSE, and volatiles) inrepresentative rejuvenated-stage lavas are consis-tent with the silicate melt metasomatism model.

[32] In general, radiogenic isotopic compositionsare also consistent with the silicate melt metaso-

matism model. Radiogenic isotopic compositionsof Hawaiian shield lavas are heterogeneous and aretypically modeled as mixtures of two main com-ponents: (1) a ‘‘KEA’’ source end-member withhigh 206Pb/204Pb and intermediate 87Sr/86Sr mostclearly represented by some historic tholeiites fromKilauea [Stille et al., 1986] and (2) a ‘‘KOO’’

Figure 11. (a and b) Lu versus Th and (c and d) Lu14 versus Th14, with format following Figure 6. Presence ofresidual garnet in the source greatly reduces Lu concentration in melts. Representative rejuvenated-stage lavas areconsistent with melting of Model 2 mantle with up to �6% garnet in the source. Most Kiekie Basalts require greaterextents melting (7% to 13%) of a more depleted source (Model 1). A few Kiekie Basalts (0.7 < Th14 < 1.5), SouthArch and T273 Honolulu Volcanics have Lu concentrations lower than melts of Model 1 mantle having garnet (3%)lherzolite mineralogy. These lavas are consistent with lower extents of melting (4 to 7%) of Model 1 mantle with upto 5% garnet. Addition of carbonatite does not significantly modify melting curves for HREE.

Figure 12. Nb/La versus Ba/Th showing (a) raw data and (b) superimposed model curves as described in Figure 6.Olivine-addition does not modify trace element ratios. Representative rejuvenated-stage lavas have significantlyhigher Nb/La and slightly higher Ba/Th consistent with melt metasomatism model. The fact that the representativerejuvenated-stage lavas have Ba/Th similar to shield lavas supports a Hawaiian plume origin for the incipient melt.High-Ba lavas (Kiekie Basalts, South Arch, T273 Honolulu Volcanics) form a trend orthogonal to that produced byvariations in partial melting of mantle metasomatized incipient silicate melts. Negative correlation between Nb/Laand Ba/Th can be produced by addition of small amounts (up to 0.2%) of carbonatite to Model 1 mantle.



16 of 34

source end-member with low 206Pb/204Pb and high87Sr/86Sr most clearly represented by alkali-poorhigh-Zr/Nb basalts and basaltic andesites of KoolauVolcano, Oahu [Frey et al., 1994; Roden et al.,1994]. The Loihi source (LOI) is intermediatebetween KEA and KOO. Posterosional and reju-venated-stage basalts require involvement of adepleted component [Tatsumoto, 1978; Stille et

al., 1986; Bizimis et al., 2005; Frey et al., 2005;Fekiacova et al., 2007]. Figure 15 shows threeseparate curves for mixing between depleted mantleand metasomatic melts having various Hawaiianend-member isotopic compositions (KOO, KEA,and LOI). Isotopic compositions of most represen-tative rejuvenated-stage lavas are consistent withour Models 1 and 2 mixing proportions.

Table 5. FTIR Analyses of Submarine Kiekie Basalt Glasses

Sample H2O (wt%) sd CO2 (ppm) sd Depth Collect (m) Depth Equilibrium (m) (XH2O)vapor

T317-R01 0.509 0.027 115 6 1691 1850 0.150T317-R02 0.459 0.015 104 6 1692 1650 0.138T317-R03 0.551 0.015 102 13 1692 1710 0.187T317-R08 0.514 0.010 119 11 1700 1910 0.148T317-R10 0.567 0.019 127 6 1657 1490 0.246T317-R13 0.440 0.021 106 10 1691 1670 0.126T317-R14 0.341 0.006 83 8 1662 1010 0.135T317-R16 0.439 0.012 185 9 1659 2750 0.077T317-R18 0.604 0.032 137 7 1673 1450 0.273T318-R01 0.363 0.015 75 5 1046 880 0.176T318-R02 0.382 0.013 73 5 1046 880 0.193T318-R03 0.373 0.012 61 5 1037 780 0.207T318-R04 0.392 0.009 83 5 1038 990 0.181T318-R06 0.401 0.011 69 5 1040 880 0.211T318-R08 0.382 0.022 106 6 1055 1210 0.142T318-R09 0.382 0.007 76 10 1054 910 0.187T318-R10 0.392 0.004 66 8 1054 820 0.218T318-R11 0.392 0.009 72 7 1037 880 0.203T318-R12 0.354 0.006 116 6 1015 1260 0.119T318-R13 0.354 0.026 89 16 991 1040 0.142T318-R14 0.325 0.008 113 9 973 1270 0.099T318-R15 0.334 0.013 135 8 1034 1480 0.090T318-R16 0.334 0.008 125 10 1050 1390 0.096T318-R17 0.344 0.009 96 4 1103 1130 0.123T322-R1 0.440 0.015 117 8 1391 1370 0.162T322-R4 0.344 0.025 127 13 1384 1630 0.084T322-R5 0.420 0.006 140 9 1385 1850 0.107T322-R30 0.325 0.011 103 12 949 1390 0.087T322-R31 0.325 0.004 86 4 947 1180 0.102T322-R32 0.315 0.034 61 7 943 870 0.131T322-R33 0.334 0.028 79 8 941 1100 0.115T323-R1 1.24 0.09 148 5 3826 3400 0.490T323-R2 1.33 0.07 140 5 3797 3600 0.531T323-R3 1.32 0.06 143 8 3787 3580 0.527T323-R4 1.13 0.04 150 6 3759 3220 0.422T323-R5 1.16 0.06 143 2 3744 3140 0.459T323-R7 1.31 0.05 164 7 3729 3860 0.481T323-R8 1.05 0.06 201 6 3713 3310 0.364T323-R9 1.22 0.03 173 6 3700 3670 0.438T323-R10 0.76 0.04 264 16 3699 3550 0.170T323-R11 0.87 0.02 294 8 3702 3870 0.211T323-R12 1.25 0.03 195 10 3664 4000 0.425T323-R15 1.12 0.06 188 12 3591 3370 0.407T323-R16 1.20 0.03 142 2 3570 3190 0.487T323-R17 1.09 0.02 146 3 3560 3110 0.402T323-R18 1.08 0.06 170 3 3550 3580 0.341T323-R19 1.15 0.01 141 4 3548 3310 0.419T323-R20 1.30 0.29 174 6 3547 4220 0.426T323-R21 1.31 0.04 179 3 3542 4320 0.424



17 of 34

5.2. Modeling Kiekie Compositions

[33] Partial melting of Model 2 mantle, however,cannot produce Kiekie compositions using reason-able extents of partial melting (Figures 6d, 7d, 8d,9d, 10d, 11d, and 12d). Model 2 mantle would

have to melt 15% to 25% to reproduce the ob-served low Th and incompatible element concen-trations in most Kiekie lavas, which is unlikely forthese alkalic compositions. Similarly, melting ofdepleted mantle would require �1% melting togenerate the Th content of Kiekie lavas. This extentof melting is too low to produce such mildly alkaliclavas. A mantle composition more depleted (withless incipient melt added) than Model 2 is required.Melting of Model 1 (99.2% depleted mantle +0.8% incipient melt) mantle provides an acceptablefit. The need for a lower proportion of incipientmelt in the Kiekie source relative to that for therepresentative rejuvenated-stage lavas is well illus-trated using the HREE (Figures 11c and 11d). Luconcentrations of the representative rejuvenated-stage lavas are bounded by melting of Model 2(2% incipient melt) mantle with 3% to 6% garnet.Lu concentrations in most Kiekie lavas cannot begenerated by melting of the Model 2 mantleregardless of the amount of garnet in the source.Nor can melting of depleted mantle reproduce theHREE concentrations. Model 1 mantle (0.8% incip-ient melt added) produces melts with appropriateHREE concentrations intermediate between meltsof depleted and Model 2 mantle compositions.Mildly alkalic Kiekie Basalts are consistent with7% to 13% batch melting of Model 1 mantle withabout 3% residual garnet. Moderately alkalic South

Figure 13. Depth of vapor saturation plotted againstdepth of collection for submarine rejuvenated-stageglasses. Depths of vapor saturation calculated accordingto Dixon [1997]. Any sample lying on the 1:1 linewould have vapor in equilibrium with the melt at thedepth of collection (eruption). Dashed lines represent±25% uncertainty. Most samples are within uncertaintyof the 1:1 line, indicating that samples were erupted inthe submarine environment close to their depth ofcollection.

Table 6. Kiekie Basalt Radiogenic Isotopic Data

Sample 208Pb/204Pb 207Pb/204Pb 206Pb/204Pb 87Sr/86Sr 2s 143Nd/144Nd 2s

SubmarineT317-R14 37.909 15.516 18.158 0.703449 10 0.513063 7T317-R18 37.773 15.454 18.102 0.703474 18 0.513042 10T318-R11 37.676 15.440 18.053 0.703444 13 0.513057 6T318-R17 37.663 15.444 18.046 0.703501 13 0.513054 6T322-R1 37.798 15.461 18.094 0.703458 15 0.513056 15T322-R32 37.857 15.514 18.148 0.703523 20 0.513059 4T323-R9 37.919 15.472 18.145 0.703466 18 0.513069 18T323-R21 37.771 15.454 18.131 0.703464 17 0.513034 9

SubaerialOX-130 37.743 15.457 18.079 0.703340 13 0.513087 1669Nii-3 37.727 15.443 18.109 0.703357 16 0.513067 969Nii-5 37.757 15.446 18.122 0.703299 35 0.512982 1569Nii-9 37.673 15.441 18.078 0.703400 11 0.512957 1270Nii-4 37.677 15.428 18.087 0.703354 20 0.513034 1070Nii-8 37.731 15.432 18.102 0.703348 21 0.513054 2270Nii-10 37.783 15.457 18.134 0.703338 33 0.513026 1870Nii-11 37.755 15.445 18.122 0.703313 47 0.513065 1670Nii-18 37.709 15.439 18.110 0.703382 7 0.513006 1870Nii-21 37.736 15.442 18.142 0.703358 2 0.512957 2270Nii-28 37.644 15.443 18.040 0.703373 6 0.513062 1575Nii-13 37.634 15.453 18.109 0.703343 12 0.513046 11



18 of 34

Arch, three high-Ba lavas from the HonoluluVolcanics, and a few Kiekie Basalts (T322R1-B,T317R10-B, T317R18-B, and some subaeriallavas) plot below the Model 1 with 3% garnetcurves. These lavas may be generated by 4 to 7%melting of garnet lherzolite with roughly 4% to 5%garnet in the source. Concentrations of high fieldstrength elements (Nb and Hf, Figures 8 and 9,respectively) in Kiekie Basalts are also consistentwith melt metasomatism models using the moredepleted Model 1 mantle.

[34] Though the more depleted source successfullypredicts HREE and HFSE element concentrationsusing reasonable extents of melting for the Kiekielavas, it considerably underestimates concentra-tions of large ion lithophile elements (Ba, Sr,LREE) and volatiles (H2O and Cl). This is most

dramatically illustrated on the plot of Nb/La versusBa/Th (Figure 12). Both addition of a metasomaticmelt and progressively lower extents of meltingcause the melt compositions of rejuvenated-stagelavas to plot above and slightly to the right of theHawaiian shield lavas. Kiekie Basalts form a trendorthogonal to that produced by magmatic processes.An additional component is therefore required toexplain that trend.

[35] With respect to radiogenic isotopes (Figure 15),there are subtle, but nonunique, differences be-tween the high-Ba and representative rejuvenated-stage lavas. The South Arch lavas have higher206Pb/204Pb than representative rejuvenated-stagelavas consistent with a larger contribution from theKEA component within the plume, whereas KiekieBasalts are shifted to higher 87Sr/86Sr and lower

Figure 14. (a and b) H2O versus Th and (c and d) H2O14 versus Th14, with format following Figure 6.Representative rejuvenated-stage lavas are consistent with partial melting of Model 2 mantle with 540 ppm H2O.Submarine Kiekie Basalts (red filled circles) correlate positively with Th consistent with other undegassedrepresentative rejuvenated-stage lavas. Kiekie T323 (red filled squares) and South Arch lavas form a vertical arraywith increasing water at relatively constant Th. No single model can explain both trends. Partial melting of Model 1mantle with 255 ppm H2O and up to 0.2% carbonatite with �10 wt% H2O can explain the H2O concentrations oflow-H2O Kiekie Basalts. High H2O concentrations in Kiekie T323 and South Arch samples require an additionalhydrous component associated with carbonatite and/or assimilation of seawater-derived component.



19 of 34

206Pb/204Pb consistent with a larger contributionfrom the KOO component.

5.3. Modeling of CarbonatiteMetasomatism

[36] We propose that carbonatite melt is the likelysource for the trace element characteristics of theBa- and Sr-enriched Kiekie Basalts. While it haslong been recognized that carbonatite metasoma-tism is a globally important process (see section 6.2below), solid evidence for a carbonatite componentin Hawaiian magmas has been lacking. Carbona-tites are characterized by enrichments in Ba, Sr,and LREE, as well as depletions in high fieldstrength elements. Carbonatite compositions arealso notoriously heterogeneous. For our modeling,we use a carbonatite trace element compositionbased on the average composition of oceaniccarbonatites from Cape Verde and the CanaryIslands [Hoernle et al., 2002]. In contrast toaverage continental carbonatites [Woolley andKempe, 1989], oceanic carbonatites are notenriched in Pb, consistent with the negative Pbanomaly observed in the Kiekie lavas. The averageoceanic carbonatite composition used here(Tables 7a, 7b, and 7c) was modified slightly to

provide an internally consistent result for allelements.

[37] Figures 6–12 show trends in lithophile traceelements produced by variable extents of meltingof Model 1 mantle to which 0.1% and 0.2%carbonatite have been added (orange dashed lines).We assume closed-system metasomatism modeledby simple binary mixing but acknowledge thatmetasomatic fluids likely interact with mantlematerials in a more complicated way. Concentra-tions of Ba, Sr, La, Nb, Hf, and Lu in Kiekie, SouthArch, and the three high-Ba Honolulu Volcanicslavas are consistent with addition of up to 0.2%carbonatite (Figures 6c–6d, 7c–7d, 8c–8d, 9c–9d,10c–10d, 11c–11d, and 12c–12d). Kiekie T323lavas can be generated by 10% to 13% batchmelting of Model 1 mantle with slightly less than0.2% carbonatite added. The positive correlationbetween Ba and Th in Kiekie T323 lavas isconsistent with trends produced by variations inpartial melting. Most high-Ba lavas can be gener-ated by 4 to 13% batch melting of Model 1 mantlewith up to 0.1% carbonatite added. Because car-bonatites are characterized by low concentrationsof high field strength elements (Nb and Ta),addition of carbonatite does not modify the con-

Figure 15. (a and b) Cl versus Th and (c and d) Cl14 versus Th14, with format following Figure 6. Representativerejuvenated-stage lavas are consistent with partial melting of Model 2 mantle with 30 ppm Cl. As in Figure 14, Kiekieand South Arch lavas form two distinct trends. Kiekie (red filled circles) correlate positively with Th consistent withvarious degrees of melting of Model 1 mantle with 13 ppm Cl and addition of up to 0.2% carbonatite with 6000 ppmCl. Kiekie T323 and South Arch lavas have higher concentrations of Cl, with the Kiekie T323 lavas forming avertical trend similar to H2O concentrations. Addition of 0.2% carbonatite with 6000 ppm Cl is insufficient to explainhigh Cl in these samples.



20 of 34

centrations of those elements. Addition of carbo-natite provides a mechanism to produce enrich-ments in Ba and La, without enrichments in Nb,thus producing the distinct negative trend on theNb/La versus Ba/Th plot (Figure 12).

[38] The volatiles require a more complicated sce-nario. The first complication is that initial volatilecontents of carbonatites are highly uncertain. Thefreshest carbonatite ever sampled is a natrocarbona-tite from Oldoinyo Lengai (BD 118 fromDawson etal. [1987] and Dawson [1989]), collected shortlyafter eruption in 1960. This sample underwentminimal modification due to alteration and con-tains 5.41 wt% H2O. However, high vesicularityindicates that eruptive volatile loss occurred, mak-ing this value a minimum estimate. Experiments

show that carbonatitic melt is capable of dissolvingsignificant quantities of water [Keppler, 2003]. At1 kbar, the water solubility reaches nearly 10 wt%,two or three times that in most silicate melts. Forour models, we arbitrarily assume a water contentof 10 wt%. This is only a rough estimate, butserves as a starting point for our models. For Cl,natural concentrations are highly variable and wearbitrarily assume a concentration of 6000 ppm.

[39] Addition of carbonatite with 10 wt% H2Oadequately explains the water concentrations inthe low-H2O Kiekie lavas, but cannot explain thehigher values in Niihau T323 and South Arch lavas(Figures 14c and 14d). No single carbonatite com-position can explain the vertical array of H2O andCl data in those lavas. Either the metasomatizingcarbonatite is heterogeneous with respect to vola-tiles or these lavas have interacted with seawater orseawater-derived brine. For South Arch lavas,addition of <0.1% carbonatite is sufficient to ex-plain the Ba, Sr, and LREE concentrations. Thiscarbonatite would have to have an associatedhydrous component to produce the observed waterenrichment. For Kiekie T323 lavas, addition of0.2% carbonatite with about 20% to 40% H2Ocould explain the water enrichment.

[40] Similar to the case for water, addition ofcarbonatite with a constant Cl content cannotexplain all the data. Using 6000 ppm Cl adequatelyexplains much of the data, but not the high-ClKiekie T323 and South Arch lavas (Figures 15cand 15d). Two possible mechanisms can explainhighly variable water and Cl concentrations in theHawaiian rejuvenated-stage lavas. Either (1) themetasomatic component is heterogeneous withrespect to volatiles or (2) the high-Ba lavas alsoassimilated seawater or a seawater-derived brineduring storage in the crust or eruption. The firstmechanism implies a common origin (carbonatitemagma with heterogeneous volatile contents),whereas the second mechanism requires a coinci-dental occurrence of two separate events (assimila-tion of brine limited to those samples metasomatizedby carbonatite).

[41] Study of microinclusions in cloudy (inclusion-rich) diamonds provides insight into mantle fluidcompositions [e.g., Navon et al., 1988; Schrauderand Navon, 1994; Izraeli et al., 2001; Klein-BenDavid et al., 2004, 2007]. Bulk compositionsof microinclusions are highly variable with threeendmember compositions: (1) hydrous-silicic meltrich in water, Si, Al, and K; (2) carbonatitic meltrich in carbonate, Mg, Ca, Fe, K, and Na; and (3)

Figure 16. Radiogenic isotopic compositions of shieldand rejuvenated-stage lavas. End-member compositionsfrom Dixon and Clague [2001, and references therein].Shield lavas are dominated by mixtures of KEA andKOO components. All rejuvenated-stage lavas areshifted toward lower 87Sr/86Sr consistent with involve-ment of a depleted component. Melts that metasomatizedepleted mantle surrounding the Hawaiian plume musthave radiogenic isotopic heterogeneity similar to that ofthe shield lavas. Radiogenic isotopic compositions ofrejuvenated-stage lavas are consistent with additionof 0.8% (Model 1) to 2% (Model 2) of incipient melt(0.5% melting) of the Hawaiian plume to averagedepleted mantle. The source of the South Arch lavaswas metasomatized by melts with KEA-like isotopicsignatures. The source of most representative rejuve-nated-stage lavas was metasomatized by melts withintermediate LOI-type compositions. The source ofhigh-Ba Kiekie lavas was metasomatized by melts withKOO-like affinity. A stronger KOO-like isotopicsignature suggests involvement of carbonatite melt fromrecycled crustal components within the plume.



21 of 34

saline fluids rich in Cl, K, and Na. All composi-tions are enriched in K and many other incompat-ible elements and are characterized by steep REEpatterns similar to those in kimberlites and lamp-roites [Schrauder et al., 1996; Raga et al., 2003].Water is always present in carbonatitic high-densityfluids in diamonds [Klein-BenDavid et al., 2007],with carbonate/(carbonate + water) molar ratiosranging from 0.5 to 0.7. Our estimated carbonatitecomposition (40 wt% CO2 and 10 wt% H2O) has acarbonate/(carbonate + water) molar ratio of 0.6,similar to carbonatitic fluids in diamonds. Chlo-rine in these carbonatitic inclusions ranges from2 to 5 wt%. The saline fluid end-member is alsowater- and chlorine-rich, containing up to 20 wt%and 49 mol%, respectively [Izraeli et al., 2001;Klein-BenDavid et al., 2007].

[42] Wyllie [1989] showed that exsolution of aH2O-CO2 vapor phase with high concentrationsof alkalis and other solutes is a normal conse-quence of crystallization of carbonated silicate

magmas or carbonatite magmas within the crust.These vapors or fluids are responsible for thecommon alteration around carbonatitic intrusions,called fenitization. More recent work [e.g., Perchuket al., 2002; Klein-BenDavid et al., 2007] modelsthe range of fluid compositions found in diamondsas the result of precipitation of alkali and divalentcarbonates from the parental carbonatitic-fluid. Theresidual fluid composition is driven into an immis-cibility gap and results in separation into saline andsilicic high density fluids. These heterogeneousfluids continue to evolve separately as they migratethrough, interact with, and fractionate within themantle, potentially leading to a diverse range ofmantle source volatile compositions. Thus, hetero-geneous volatile contents in Kiekie Basalts affectedby carbonatite metasomatism could easily resultfrom heterogeneous fluids associated with carbo-natite melts.

[43] Addition of a seawater-derived brine can alsoenrich lavas in H2O and Cl without modifying

Table 7a. Mineral and Bulk Distribution Coefficientsa

Olivine Opx Low-Ca Cpx Garnet Sulfide Adj.Hawaiian

Mantle 1% gtHawaiian

Mantle 3% gt

Rb 0.0003 0.0002 0.0004 0.0002 - 0.00032 0.00032Ba 0.000005 0.000006 0.0004 0.00007 - y-g 0.00014 0.00014CO2 * 0.0002 0.0002Th 0.00005 0.002 0.0057 0.009 - 0.0022 0.0024Cl * 0.004 0.0042Nb 0.0005 0.004 0.010 0.015 - 0.0042 0.0045Ta 0.0005 0.004 0.010 0.015 - 0.0042 0.0045La 0.0005 0.004 0.015 0.007 - y 0.0085 0.0085H2O 0.0005 0.004 0.02 - y 0.0097 0.0097Ce 0.0005 0.004 0.038 0.017 - y 0.014 0.014Pb 0.003 0.009 0.009 0.005 150 0.014 0.014Sr 0.00004 0.0007 0.091 0.0007 - 0.031 0.031Nd 0.00042 0.012 0.088 0.064 - 0.032 0.033Sm 0.0011 0.02 0.151 0.23 - y 0.047 0.052Zr* - * 0.028 0.032Hf 0.0022 0.03 0.14 0.40 - y 0.041 0.048Eu* - * 0.045 0.055Gd 0.004 0.03 0.16 1.2 - y 0.0614 0.085Tb* - * 0.068 0.085Dy 0.008 0.035 0.17 2.0 - 0.085 0.125Y* - * 0.10 0.130Ho* - * 0.102 0.140Er 0.013 0.045 0.18 3.0 - 0.102 0.162Tm* - * 0.115 0.175Yb 0.03 0.08 0.25 4.0 - 0.148 0.228Lu 0.02 0.12 0.2759 5.0 - 0.165 0.264

aBulk distribution coefficients for various mineral assemblages were calculated using the above mineral distribution coefficients from Salters et

al. [2002], Hart and Gaetani [2006] for Pb, and Aubaud et al. [2004] for H2O. Asterisks indicate interpolated bulk distribution coefficients. Mineralproportions for Hawaiian mantle based on melt reactions of Salters et al. [2002]. Hawaii Mantle 1% gt = garnet lherzolite composed of 57% olivine,8% orthopyroxene, 34% low-Ca clinopyroxene, and 1% garnet. Hawaii Mantle 3% gt = garnet lherzolite composed of 55% olivine, 8%orthopyroxene, 34% low-Ca clinopyroxene, and 3% garnet. Pb distribution coefficients assume 0.015% sulfide in the source. Elements with a ‘‘y’’have had their bulk distribution coefficients adjusted (Adj.) to provide a self-consistent model for generation of Hawaiian tholeiitic lavas.



22 of 34

lithophile element concentrations. Assimilation ofa seawater-derived brine usually occurs in systemswith long-lived hydrothermal systems, as is foundat fast spreading centers and Loihi Seamount[Michael and Schilling, 1989; Jambon et al.,1995; Michael and Cornell, 1998; Kent et al.,1999; Dixon and Clague, 2001]. Assimilation ofa pure seawater contamination, possibly throughinteraction with serpentine, at slow-spreading cen-ters or at small oceanic islands typically affectswater, but not Cl [Simons et al., 2002]. We do notfind evidence for sustained hydrothermal systemsassociated with the submarine Kiekie vents. How-ever, the submarine Kiekie Basalts erupted in flat-topped submarine cones [Clague and Reynolds,2000], which may have provided more opportunity

for foundering of the lava lake crust and interactionof magma with altered materials. Though wecannot rule out contamination by seawater as thesource of the high water and Cl concentrations inKiekie T323 and South Arch lavas, we prefer amodel in which carbonatite melt differentiatesduring flow through the mantle to produce hetero-geneous fluid compositions as postulated for theorigin of heterogeneous high-density fluids in dia-monds [e.g., Perchuk et al., 2002; Klein-BenDavidet al., 2007].

5.4. Composition of MetasomaticCarbonatite Derived From Hawaiian Plume

[44] The primitive mantle-normalized trace ele-ment composition of the carbonatite required toproduce the trace element composition of the high-Ba lavas is plotted on Figure 16a. In general, thecomposition of the carbonatite required by ourmodeling matches the overall primitive mantlenormalized pattern of oceanic carbonatite withenrichments in Ba, Sr, LREE, and volatiles andrelatively low concentrations of Nb, Ta, Pb, Zr, Hf,and HREE. All trace element concentrations liewithin the minimum and maximum concentrationsfound in oceanic carbonatites from Cape Verde andCanary Islands [Hoernle et al., 2002]. In order foraddition of up to 0.2% carbonatite to explain thecompositional variation in all high-Ba lavas, themodel carbonatite must have higher Sr and Baconcentrations but lower Th, U, LREE, Zr, and Hfthan the average oceanic carbonatite composition.Note that the enrichments of H2O and Cl (10 wt%and 6000 ppm, respectively) relative to primitivemantle are similar to those observed for LREE.

5.5. Forward Model of Rejuvenated-StageMagma Compositions

[45] Primitive mantle normalized trace elementcompositions of lavas predicted by the melt meta-somatism and melt plus carbonatite metasomatismmodels are shown on Figure 17. Representativerejuvenated-stage lava Kauai-Oahu 4D can begenerated by 7% melting of a Model 2 mantle(98% depleted mantle + 2% incipient melt) usinggarnet lherzolite mineralogy.

[46] Kiekie T323-R10 and -R03 represent thecompositions of volatile- and Ba-rich lavas. Melt-ing (10%) of Model 1 mantle (99.2% depletedmantle + 0.8% incipient melt) successfully predictsthe Nb, Ta, Zr, Hf, Yb, and Lu concentrations butsignificantly underestimates the Ba, Sr, LREE, andvolatile concentrations. Melting (10%) of Model 1

Table 7b. Bulk Distribution Coefficients for Refrac-tory Mantle Compositiona

RefractoryMantle3% gt

RSVmantle3% gt

RSVmantle6% gt

RSVmantle3% gt +

phlog + ox

Rb 0.0003 0.003 0.003 0.062Ba 0.00005 0.00006 0.00006 0.011CO2 0.0002 0.0002 0.0002Th 0.0011 0.0012 0.00145 0.00144Cl 0.0028 0.0028 0.0028Nb 0.0022 0.0028 0.0028 0.023Ta 0.0022 0.0028 0.0028 0.033La 0.0023 0.005 0.005 0.006H2O 0.0051 0.0057 0.0058Ce 0.0053 0.0061 0.0066Pb 0.028 0.028Sr 0.010 0.007 0.009 0.012Nd 0.013 0.012 0.0165Sm 0.026 0.023 0.036Zr* 0.05 0.060Hf 0.031 0.04 0.052 0.08Eu* 0.033 0.045 0.055Gd 0.059 0.042 0.060Tb* 0.05 0.068Dy 0.087 0.07 0.085Y* 0.11 0.13Ho* 0.12 0.15Er 0.12 0.12 0.20Tm* 0.15 0.270Yb 0.18 0.18 0.300Lu 0.20 0.21 0.359

aRefractory mantle modal proportions: 80% olivine, 6% orthopyr-

oxene, 11% low-Ca clinopyroxene, and 3% garnet. Bulk distributioncoefficients for the RSV mantle were estimated to fit rejuvenated-stagelava trace element data and are similar to those for a refractory(pyroxene-poor) modal mineralogy. Bulk distribution coefficients forRSV mantle with accessory phlogopite and an oxide phase such astitanite were estimated to fit the basanitic and nephelinitic rejuvenated-stage lava compositions. Proportions of residual accessory phases werenot estimated.



23 of 34

mantle to which 0.2% carbonatite has been addedsuccessfully predicts all trace element concentra-tions with the exception of H2O and Cl. LREEconcentrations in Kiekie T323-R10 suggest a largercarbonatite contribution than in Kiekie T323-R03;however, both H2O and Cl in R10 are lower than inR03, implying decoupling of volatiles from othertrace elements as discussed above.

6. Discussion

6.1. Source of Carbonate in HawaiianPlume

[47] Recycling of crustal materials into the deepmantle by subduction is an important source ofheterogeneity in the Earth’s mantle. Recent studieshave shown that water is efficiently dehydratedfrom the downgoing slab during subduction [Dixonand Clague, 2001; Dixon et al., 2002; Rupke et al.,2006]. Carbonate, however, is widely thought to

mostly survive subduction-related dehydration andsilicate melting [Yaxley and Green, 1994; Molinaand Poli, 2000; Kerrick and Connolly, 2001;Dasgupta et al., 2004; Brenker et al., 2007].Recent studies have shown that transport of CO2

into the deep mantle through subduction and crust-al recycling is likely [e.g., Molina and Poli, 2000;Kerrick and Connolly, 2001; Dasgupta et al.,2004]. Molina and Poli [2000] present experimen-tal results that show that amphibole coexists withcalcite at �1.4 Gpa, with dolomite at 1.4 � P �1.8 Gpa, and with dolomite+magnesite at pressureshigher than 1.8 Gpa. Significant decarbonation ofsubducted crust is feasible only at low pressures (inthe forearc region) and at relatively high temper-atures (e.g., young oceanic crust subducted at slowconvergence rates) [Molina and Poli, 2000]. Oncestable subduction conditions have been reached,CO2 is fractionated into the solid phase and deeprecycling of CO2 is expected. Kerrick and Connolly[2001] present a subduction zone devolatilization

Table 7c. Mantle End-Member Trace Element Compositionsa

DMM HWM0.5% Meltof HWM Model 1 Model 2 Carbonatite

Model 1 +0.1% Carb

Model 1 +0.2% Carb

Rb 0.020 0.40 75.2 0.62 1.52 0.58 0.62 0.62Ba 0.227 5 973 8.01 19.7 6000 14.0 20.0CO2 50 500 96200 819 1970 400000 1220 1620Th 0.004 0.04 5.43 0.047 0.112 3.35 0.051 0.054Cl 1.3 13 1420 12.6 29.6 6000 18.6 24.6Nb 0.0864 0.77 81.7 0.74 1.72 35 0.77 0.81Ta 0.0056 0.05 5.31 0.048 0.11 0.050 0.048 0.048La 0.134 0.79 58.7 0.60 1.31 250 0.85 1.10H2O 65 350 23800 255 540 100000 355 455Ce 0.421 2.00 106 1.26 2.52 390 1.65 2.04Pb 0.014 0.085 2.58 0.035 0.065 9.93 0.044 0.054Sr 6.09 27 752 12.1 21.0 13000 25.1 38.0Nd 0.483 1.7 44.7 0.84 1.37 270 1.11 1.38Sm 0.21 0.57 10.0 0.29 0.41 50 0.34 0.39Zr 3.5 12.5 339 6.19 10.2 3.41 6.18 6.18Hf 0.127 0.37 7.03 0.18 0.27 0.05 0.18 0.18Eu 0.086 0.19 3.18 0.11 0.15 6.43 0.12 0.12Gd 0.324 0.69 7.72 0.38 0.47 23.0 0.41 0.43Tb 0.064 0.11 1.23 0.073 0.087 2.6 0.076 0.078Dy 0.471 0.75 5.80 0.51 0.58 14.6 0.53 0.54Y 3.13 4.4 32.8 3.37 3.72 87.7 3.45 3.53Ho 0.108 0.152 1.05 0.12 0.13 3.6 0.12 0.12Er 0.329 0.43 2.59 0.35 0.37 9.2 0.36 0.37Tm 0.045 0.059 0.33 0.047 0.051 0.90 0.048 0.049Yb 0.348 0.44 1.90 0.36 0.38 4.0 0.36 0.37Lu 0.056 0.070 0.26 0.058 0.060 0.50 0.058 0.059

aAll values in ppm by weight. Abbreviations: DMM is depleted MORB mantle from Workman and Hart [2005]. HWM is average Hawaiian

plume mantle calculated in this study to fit Hawaiian shield lavas by 4–15% batch melting using bulk distribution coefficients for garnet lherzolitewith 1 to 3% garnet. Incipient (0.5%) melt of Hawaiian plume mantle calculated using garnet (3%) lherzolite distribution coefficients. Model 1 is99.2% depleted mantle +0.8% of 0.5% partial melt of Hawaiian plume. Model 2 is 98% depleted mantle +2% of 0.5% partial melt of Hawaiianplume. Carbonatite composition calculated to give compositions of high-Ba lavas by addition of 0.2% carbonatite to model 1 mantle. Resultingcarbonatite composition is similar to average oceanic carbonatite [Hoernle et al., 2002].



24 of 34

Figure 17. (a) Primitive mantle normalized trace element composition of oceanic carbonatites from Cape Verde andCanary Islands [Hoernle et al., 2002] and Hawaiian carbonatite composition. The Hawaiian carbonatite compositionrequired to produce compositional variations observed in Kiekie lavas is similar to that of oceanic carbonatites. (b)Representative rejuvenated-stage lavas (e.g. Kauai-Oahu Channel 4D) can be generated by partial melting (7%) ofModel 2 mantle using garnet (4.5%) lherzolite mineralogy. (c) Kiekie lavas cannot be generated by partial melting ofsilicate melt metasomatized mantle. Ten percent melting of mantle model 1 using garnet (3%) lherzolite mineralogyreproduces Th, HFSE, and HREE concentrations of Kiekie lavas but significantly underestimates Ba, Sr, LREE, H2O,and Cl concentrations. Ten percent melting of Model 1 mantle, metasomatized by 0.2% carbonatite, predicts theconcentrations of all nonvolatile trace elements. H2O and Cl concentrations in Kiekie lavas require addition of ametasomatic component heterogenous with respect to volatiles.



25 of 34

model for pressures up to 6 Gpa (180 km) based onphase equilibria computed for an average oceanicmetabasalt. They show that decarbonation is neg-ligible along cold and intermediate geotherms andlimited along high temperature geotherms. Thus,given that the Hawaiian plume is thought to con-tain recycled crustal components, it is reasonableto expect elevated carbonate contents in thosecomponents.

[48] Evidence for a recycled crustal component isstrongest in lavas collected from the Koolau shieldon Oahu. These lavas define an extreme endmem-ber [e.g., Frey et al., 1994; Roden et al., 1994;Lassiter and Hauri, 1998; Blichert-Toft et al.,1999; Jackson et al., 1999; Huang and Frey,2003, 2005] characterized by high 87Sr/86Sr,187Os/188Os and d18O and by low 143Nd/144Nd,176Hf/177Hf, and 206Pb/204Pb. In terms of major andtrace elements, the Koolau lavas have relativelyhigh SiO2 content, SiO2/total iron, Al2O3/CaO, La/Nb, Sr/Nb, and low total iron and CaO contents.These geochemical characteristics have been inter-preted as evidence for recycled oceanic crust,including sediments, in the source of Koolau lavas.Huang and Frey [2005] argue that the distinctivegeochemical characteristics of the Koolau lavascan be explained by mixing between high-temper-ature melts derived from garnet peridotite with<20% dacitic melt derived from garnet pyroxenite(recycled oceanic crust). Huang and Frey [2005]also suggest that the recycled crustal componentcontains carbonate, <3% of either ancient recycledphosphate-bearing carbonate or perhaps sedimentwith an abundant hydrothermal component. Thus,it is likely that the Koolau component within theHawaiian plume contains the highest carbon con-centrations that could be remobilized in the form ofcarbonatite melts, consistent with the KOO-likeradiogenic isotopic compositions of the Kiekielavas.

6.2. Evidence for CarbonatiteMetasomatism in Oceanic Environments

[49] Metasomatism by either silicate and/or carbo-natitic melts or H2O- or CO2-rich fluids has beenrecognized in upper mantle rocks worldwide [e.g.,Menzies and Hawkesworth, 1987; Dautria et al.,1992; Ionov et al., 1993; Rudnick et al., 1993; LeRoex and Lanyon, 1998; Yaxley et al., 1998; Yaxleyand Brey, 2004; Tappe et al., 2006, 2007]. Theimportance of CO2 in the generation of stronglyalkalic magmas is supported by both the commonfield association of alkaline and carbonatitic mag-

mas [e.g., Allegre et al., 1971; Bell, 1998; Hoernleet al., 2002; Schultz et al., 2004; Tappe et al.,2007] and results of experimental studies [e.g.,Eggler and Mysen, 1976; Eggler, 1978; Hirose,1997; Dasgupta et al., 2005, 2006, 2007]. Thoughmore commonly found in continental environ-ments, carbonatite metasomatism in the oceanicmantle has been proposed by various researchersto explain (1) the presence of carbonate and/ormicrodiamonds in xenoliths [e.g., Kogarko et al.,1995; Frezzotti et al., 2002; Wirth and Rocholl,2003]; (2) trace element compositions of secondaryminerals within peridotite xenoliths [e.g., Hauri etal., 1993; Salters and Zindler, 1995]; (3) thepresence of incipient alkali silicate or carbonatiticglasses within peridotitic xenoliths [Hauri et al.,1993; Schiano et al., 1994; Schiano and Clocchiatti,1994; Sen et al., 1996; Wulff-Pedersen et al., 1996;Neumann and Wulff-Pedersen, 1997; Vannucci etal., 1998;Coltorti et al., 1999;Wulff-Pedersen et al.,1999; Keshav and Sen, 2003]; and (4) trace ele-ment trends in some posterosional lavas fromKauai [Maaløe et al., 1992] and in three phlogopitenephelinites collected on the Hilina bench off ofKilauea [Sisson et al., 2002]. In addition, carbo-natite melts have been erupted in the Cape VerdeIslands [Silva et al., 1981] and in the CanaryIslands [Hoernle et al., 2002]. However, prior tothis study, there has been no solid evidence that acarbonatite component has directly influencederuptive products in Hawaii.

6.3. Experimental Basis for Generation ofCarbonatites in Upwelling Mantle Plumes

[50] Our model of silicate melt and carbonatitemetasomatism surrounding the Hawaiian plume isconsistent with models of melting and metasoma-tism in upwelling carbonated oceanic mantle [e.g.,Wyllie, 1988;Wyllie and Ryabchikov, 2000; Presnalland Gudfinnsson, 2005; Dasgupta et al., 2005,2006, 2007]. Wyllie [1988] first presented a meltingmodel for the Hawaiian plume, which predictedformation of trace amounts of volatile-chargedincipient melt at depths of 350 to 150 km as theplume crosses the solidus for peridotite-C-H-O. Hepredicted these incipient melts would be swampedby major melting at the center of the plume butmight be detectable on the margins of the plume.This and subsequent models provided an importantframework for the spatial and temporal variation inHawaiian magmatism as a function of temperaturevariation but did not include lithologic heterogene-ity, such as the presence of carbonated eclogite,within the plume.



26 of 34

[51] Experimental data at 1.5 to 5.5 GPa show thatmelting of carbonated peridotite and eclogite pro-duces carbonatite melts [Falloon and Green, 1990;Yaxley and Green, 1994; Dalton and Presnall,1998; Hammouda, 2003; Yaxley and Brey, 2004].New melting experiments [Dasgupta et al., 2005,2006, 2007] show that melting of carbonatedeclogite and lherzolite at progressively higher tem-peratures produces carbonatite melt at roughly300–400�C below their CO2-free solidus, followedby formation of strongly silica-undersaturated sil-icate melt at 120 to 150�C below the CO2-freesolidus. Carbonate and silicate melts coexist over awide range of temperatures during partial meltingof carbonated eclogite and lherzolite at 3 GPa.Thus, recycled carbonates or carbonated eclogite/peridotite incorporated in upwelling mantle mayproduce carbonatite melt at depths of �410 km inthe plume center and at �300 km along the plumemargin [Dasgupta et al., 2004, 2006, 2007]. Smallvolume, highly mobile carbonatite melts releasedfrom carbonated eclogite/peridotite at these depthscould be an effective metasomatic agent for sur-rounding peridotite and would be expected tomigrate greater distances from the hot plume corethan incipient silicate melts because of their greatermobility [Minarik and Watson, 1995]. Productionof carbonatite at these depths favors metasomatismof depleted mantle thermally entrained by theplume, rather than shallow lithosphere, consistentwith recent Pb-isotopic studies [Bizimis et al.,2005; Frey et al., 2005; Fekiacova et al., 2007].

[52] Dasgupta et al. [2006] propose that melting ofa heterogeneous carbonated mantle involves re-peated melting of heterogeneities, metasomatismof surrounding, more refractory peridotite, followedby partial melting of the metasomatized region, withcorrespondingly large effects on highly incompati-ble trace elements. This model provides a goodexplanation for compositional variations observedin Hawaiian rejuvenated-stage lavas. Carbonated

eclogite within the Hawaiian plume is the first tomelt during plume ascent. Carbonatitic melt meta-somatizes the plume itself and depleted astheno-sphere thermally entrained on the margins of theHawaiian plume. At lower pressures, silica-under-saturated silicate melts are also produced fromcarbonated eclogite/lherzolite and contribute tothe metasomatic signature. At the plume center,the metasomatic component is drowned out by thehigh-extent melting of the lherzolitic plume mate-rial that produces tholeiitic shield lavas. The meta-somatic component is best preserved at the marginsof the plume, where low extents of melting of themetasomatized depleted mantle surrounding theplume are sampled during flexural uplift.

[53] Similar multistage metasomatic models havebeen used to explain the occurrence of carbonatitesand alkalic magmas other locations, includingAillik Bay, Labrador [Tappe et al., 2006, 2007],the In Teria district in the Sahara Basin [Dautria etal., 1992], the Himalayan collision zone in westernSichauan, SW China [Hou et al., 2006], andnorthwestern Namibia [Le Roex and Lanyon,1998], indicating that this is a significant processworld-wide.

6.4. Implications for Melt GenerationModels

[54] An alternate model proposed for generation ofrejuvenated-stage lavas involves progressive deple-tion of heterogeneous mantle during multistagemelting [Bianco et al., 2005]. In their model,‘‘primary’’ melting to produce shield lavas depletesthe mantle in more enriched and fusible pyroxeniteand hydrous peridotitic components. This depletedmantle is laterally advected beneath the flexuralarch, where is undergoes a second stage of decom-pression melting. Though we agree with their over-all physical model for generating rejuvenated-stagelavas, their geochemical model cannot explain theKiekie Basalt compositions. First, progressivemelting of a heterogeneous mantle cannot producethe observed enrichments in Ba, Sr, LREE, andvolatiles relative to Th and HFSE. Second, theirmodel predicts significantly lower extents of partialmelting beneath the flexural arch compared to thatwithin the plume center. Though this is true for somerejuvenated-stage lavas, Kiekie Basalts (4–13%melting) are produced by extents of melting com-parable that of many shield lavas (4–15%).The easiest way to increase extents of melting at agiven temperature is to increase the mantle volatile

Table 8. Volatile Contents in Modeled Mantle End-Members

H2O (ppm) CO2 (ppm)

Depleted mantle 65 50Hawaiian mantle 350 500Model 1 255 820Model 1 + 0.1% carbonatite 355 1220Model 1 + 0.2% carbonatite 455 1620Model 2 540 1970



27 of 34

content. The volatile contents of our various mantlecompositions are shown in Table 8.

[55] The water content in our Hawaiian plumemantle (350 ppm) is within error of the value(450 ± 190 ppm) determined by Wallace [1998]based on undegassed Kilauea lavas. Our metasom-atized sources have water contents slightly lessthan to greater than the Hawaiian plume end-member. The water content for Model 2 mantlesource (540 ppm) for the representative rejuvenated-stage lavas is identical to the value (525 ± 75 ppm)previously determined for the source of the NorthArch volcanics [Dixon et al., 1997]. Our estimatedmantle CO2 concentrations (0.08 to 0.2 wt%) areconsistent with the results of Dasgupta et al.[2007], who state ‘‘the characteristic enrichmentsin CaO and depletions in SiO2 of nephelinites andmelilitites may be matched by small-degree (1–5%) partial melts of peridotite with 0.1 to 0.25%wt% bulk CO2.’’ Thus for a given amount ofdecompression, our metasomatized mantle compo-sitions will begin melting at lower temperaturesand produce more melt than volatile-free depletedmantle compositions predicted by progressive de-pletion models. In fact, metasomatic enrichment involatiles may be required to produce melt duringlithospheric flexure decompression.

6.5. Plume ‘‘Degassing’’: Decoupling ofVolatile and Lithophile Tracers

[56] Several studies have documented decouplingof noble gas and radiogenic isotopic tracers withinand surrounding the Hawaiian plume. First, thehighest 3He/4He ratios in Hawaiian lavas are ob-served at Loihi [e.g., Rison and Craig, 1983; Kurzet al., 1983, 1987; Honda et al., 1993; Valbracht etal., 1996; Kaneoka et al., 2002], spatially offsetfrom the maximum in radiogenic isotopic tracersdefined by the Mauna Kea (KEA) and Koolau(KOO) end-members. Second, lavas from theSouth Arch, 200 km upstream from the plumecenter, have elevated 3He/4He, in contrast to theMORB-like values found in other rejuvenated-stage lavas [Dixon and Clague, 2001; Hanyu etal., 2005]. Valbracht et al. [1996] first proposedthat plume ‘‘degassing’’ was responsible for thedecoupling of noble gas and radiogenic isotopictracers within the Hawaiian plume. They proposedthat incipient CO2-dominated melts form at rela-tively shallow depths near the base of the litho-sphere, possibly when the plume begins to flex bydrag of the overriding lithosphere. They arguedthat noble gases preferentially partition into thesegas-dominated melts, which then permeate and

metasomatize the lithosphere. The residual plumemantle will be strongly depleted in noble gases butnot in lithophile tracers. Schilling et al. [1999]propose a similar mechanism (defluidization) toexplain the longer wavelength of the 3He/4Hegradient relative to that of Sr-Nd-Pb isotopic ratiosalong the Kolbeinsey Ridge north of the Icelandplume. Formation of a CO2-vapor phase at thesedepths (>100 km), however, is unlikely. Dixon[1997] showed that exsolution of a CO2-rich fluidphase for a range of alkalic to tholeiitic basaltsdoes not begin until much shallower depths(�20–50 km), well within the lithosphere.

[57] Dixon and Clague [2001] proposed that whilemetasomatism occurs everywhere surrounding theHawaiian plume, a distinctive hydrous melt de-rived from the plume is required in advance of theplume, in order to explain the distinct KEA-likeradiogenic isotopic compositions along with ele-vated 3He/4He and water contents in South Archlavas compared to other rejuvenated-stage lavas.Hanyu et al. [2005] builds on this model, suggest-ing a temporal variation of incipient melt compo-sitions. They proposed the earliest formed incipientmelts of the Hawaiian plume will be volatile-richand will metasomatize the asthenosphere-litho-sphere boundary upstream of the plume. Incipientmelts produced downstream from the plume willbe volatile-poor. Hanyu et al. [2005] suggest thatthe South Arch taps the volatile-rich metasomat-ized zone characterized by high water and3He/4He, while other rejuvenated and submarinealkalic lavas are derived from the volatile-poormetasomatized zone characterized by low waterand MORB-like 3He/4He. Kiekie Basalts do notsupport Hanyu et al.’s model because they arevolatile-rich but yet have MORB-like 3He/4He[Hanyu et al., 2005].

[58] By adding lithologic heterogeneity to this mod-el, we can reconcile this seeming inconsistency.Melting of carbonated eclogite (KOO component)may produce volatile-rich carbonatite compositionswith low 3He/4He, whereas melting of carbonatedperidotite (KEA component) may produce carbo-natites with higher 3He/4He. Thus, heterogeneity inthe composition of metasomatizing melts mayreflect both the lithology of the plume compo-nent being melted and the location of melting(early versus late). In contrast to Valbracht et al.[1996] and Hanyu et al. [2005], we suggest thatincipient melting of the Hawaiian plume beginsas deep as 400 km, where melting of carbonatedeclogite/lherzolite within the plume produces



28 of 34

both carbonatite and silicate incipient melts. Deepmetasomatism will allow the metasomatic signatureto be transported away from the plume by bothdiffusion (percolation) and plume-driven advection.

6.6. Implications for Seismic Low VelocityZone

[59] It has been proposed that low seismic veloc-ities near the top of the mantle transition zone(�410 km depth) [Revenaugh and Sipkin, 1994;Vinnik and Farra, 2002; Song et al., 2004] repre-sent the effects of high water content [e.g., Berco-vici and Karato, 2003]. Laboratory experimentsthat produce hydrous phases stable at these pres-sures [e.g., Kohlstedt et al., 1996; Smyth et al.,1997] are frequently cited to support this hypoth-esis. However, a wet transition zone is inconsistentwith geochemical data and models suggestingefficient dehydration of subducted lithosphereresulting in a net flux of water out of the mantleinto the exosphere [Dixon and Clague, 2001;Dixon et al., 2002; Rupke et al., 2006]. Carbonatitemelts, not hydrous melts or minerals, may accountfor these slow seismic velocities [Dasgupta et al.,2004; Presnall and Gudfinnsson, 2005].

7. Conclusions

[60] 1. Trace element concentrations of most reju-venated-stage lavas can be generated by smalldegrees (1% to 9%) of melting of depleted perido-tite recently metasomatized by a few percent of anenriched incipient melt (�0.5% melting) from theHawaiian plume.

[61] 2. Kiekie lavas from Niihau differ from otherrejuvenated-stage lavas in that they show overalldepletion in incompatible elements combined withvariable degrees of Ba, Sr, LREE, H2O, and Clenrichment.

[62] 3. Compositions of Kiekie Basalts are bestexplained by partial melting of a depleted perido-tite source recently metasomatized by up to 0.2%carbonatite, in addition to small amounts of incip-ient silicate melt of the Hawaiian plume.

[63] 4. The carbonatite metasomatic component issimilar in composition to oceanic carbonatites fromCape Verde and the Canary Islands.

[64] 5. We prefer a model whereby heterogeneousfluids, similar to those observed as microinclusionsin diamonds, are produced as the carbonatite infil-trates and interacts with the surrounding peridotite

producing elevated H2O and Cl contents. Wecannot rule out assimilation of seawater or aseawater-derived brine to explain the extreme en-richment in H2O and Cl in Kiekie and South Archlavas.

[65] 6. Our model is consistent with predictionsbased on carbonated eclogite and peridotite meltingexperiments in which multistage metasomatism ofthe surrounding asthenosphere by carbonatite andsilica-undersaturated silicate melts is a naturalconsequence of melting of heterogeneous litholo-gies within the Hawaiian plume.

[66] 7. The metasomatic component is best pre-served at the margins of the plume, where lowextents of melting of the metasomatized depletedmantle surrounding the plume are sampled duringflexural uplift.

[67] 8. Partitioning of He into incipient carbonatiteand silicate melts formed during plume ascentstarting as deep as �400 km provides a mechanismto decouple He from lithophile tracers of plumeend-members.

[68] 9. Formation of incipient melts at depths of100 to 400 km caused by carbon-related reductionin solidus temperatures of eclogitic and peridotiticmaterials should be considered as an alternate towet transition zone models.

Acknowledgments

[69] We thank Captain Ian Young and the crew of the R/V

Western Flyer and Chief Pilot Dale Graves and the ROV

Tiburon crew. We also thank the rest of the scientific party on

the ship (Alicia Davis, Ken Hon, Jenny Paduan, Jennifer

Reynolds) who ably assisted in completing all the dives and

describing and curating the many samples recovered. E. D.

Jackson, G. Brent Dalrymple, Melvin Beeson, Richard Doell,

and DAC collected the samples from Niihau Island with the

permission and assistance of the Robinson family. We thank

Rajdeep Dasgupta and Michael Bizimis for their thoughtful

reviews. DAC was supported by grants from the David and

Lucile Packard Foundation to MBARI. JED was supported by

NSF-OCE351149.

References

Allegre, C. J., F. Pineau, M. Bernat, and M. Javoy (1971),Evidence for the occurrence of carbonatites on the CapeVerde and Canary islands, Nature, 233, 103–104.

Aubaud, C., E. H. Hauri, and M. M. Hirschmann (2004),Hydrogen partition coefficients between nominally anhy-drous minerals and basaltic melts, Geophys. Res. Lett., 31,L20611, doi:10.1029/2004GL021341.

Bell, K. (1998), Radiogenic isotope constraints on relation-ships between carbonatites and associated silicate rocks –



29 of 34

a brief review, J. Petrol., 39, 1987–1996, doi:10.1093/petrology/39.11.1987.

Bercovici, D., and S.-I. Karato (2003), Whole-mantle convec-tion and the transition-zone water filter, Nature, 425, 39–44,doi:10.1038/nature01918.

Bergmanis, E. C., J. M. Sinton, and F. A. Trusdell (2000),Rejuvenated volcanism along the southwest rift zone, EastMaui, Hawaii, Bull. Volcanol., 62, 239–255, doi:10.1007/s004450000091.

Bianco, T. A., G. Ito, J. M. Becker, and M. O. Garcia (2005),Secondary Hawaiian volcanism formed by flexural arch de-compression, Geochem. Geophys. Geosyst., 6, Q08009,doi:10.1029/2005GC000945.

Bizimis, M., G. Sen, and V. J. M. Salters (2005), Hf-Nd-Srisotope systematics of garnet pyroxenites from Salt LakeCrater, Oahu, Hawaii: evidence for a depleted componentin Hawaiian volcanism, Geochim. Cosmochim. Acta, 69,2629–2646, doi:10.1016/j.gca.2005.01.005.

Blichert-Toft, J., F. A. Frey, and F. Albarede (1999), Hf isotopeevidence for pelagic sediments in the source of Hawaiianbasalts, Science, 285, 879–882, doi:10.1126/science.285.5429.879.

Brenker, F. E., C. Vollmer, L. Vincze, B. Vekemans, A. Szymanski,K. Janssens, I. Szaloki, L. Nasdala, W. Joswig, and F. Kaminsky(2007), Carbonates from the lower part of the transition zone oreven the lower mantle, Earth Planet. Sci. Lett., 260, 1–9,doi:10.1016/j.epsl.2007.02.038.

Chen, C.-Y., and F. A. Frey (1983), Origins of Hawaiian tho-leiite and alkalic basalt, Nature, 302, 785–789, doi:10.1038/302785a0.

Chen, C.-Y., and F. A. Frey (1985), Trace element and isotopicgeochemistry of lavas from Haleakala volcano, East Maui,Hawaii: Implications for the origin of Hawaiian lavas,J. Geophys. Res., 90 , 8743 – 8768, doi:10.1029/JB090iB10p08743.

Chen, C.-Y., F. A. Frey, and M. O. Garcia (1990), Evolution ofalkalic lavas at Haleakala Volcano, east Maui, Hawaii, Con-trib. Mineral. Petrol., 105, 197 – 218, doi:10.1007/BF00678986.

Clague, D. A. (1987), Hawaiian alkaline volcanism, in Alka-line Igneous Rocks, edited by J. G. Fitton and B. G. J. Upton,Geol. Soc. London Spec. Publ., 30, 227–252.

Clague, D. A., and G. B. Dalrymple (1987), The Hawaiian-Emperor volcanic chain: Part 1. Geologic evolution, U. S.Geol. Surv. Prof. Pap., 1350, 5–54.

Clague, D. A., and G. B. Dalrymple (1988), Age and petrologyof alkalic postshield and rejuvenated-state lava from Kauai,Hawaii, Contrib. Mineral. Petrol., 99, 202 – 218,doi:10.1007/BF00371461.

Clague, D. A., and G. B. Dalrymple (1989), The tectonic andgeologic setting of the Hawaiian-Emperor volcanic chain,in The Eastern Pacific Ocean and Hawaii, Geol. of N. Am.,vol. N, edited by E. L. Winterer, D. M. Hussong, and R. W.Decker, pp. 188–217, Geol. Soc. of Am., Boulder, Colo.

Clague, D. A., and F. A. Frey (1982), Petrology and traceelement geochemistry of the Honolulu Volcanics, Oahu: Im-plications for the oceanic mantle below Hawaii, J. Petrol.,23, 447–504.

Clague, D. A., and J. G. Moore (2002), The proximal part ofthe giant submarine Wailau landslide, Molokai, Hawaii, J.Volcanol. Geotherm. Res., 113, 259–287, doi:10.1016/S0377-0273(01)00261-X.

Clague, D. A., and J. Reynolds (2000), Formation of submarineflat-topped volcanic cones in Hawaii, Bull. Volcanol., 62,214–233, doi:10.1007/s004450000088.

Clague, D. A., R. T. Holcomb, J. M. Sinton, R. S. Detrick, andM. E. Torresan (1990), Pliocene and Pleistocene alkalicflood basalts on the seafloor north of the Hawaiian islands,Earth Planet. Sci. Lett., 98, 175–191, doi:10.1016/0012-821X(90)90058-6.

Clague, D. A., W. S. Weber, and J. E. Dixon (1991), Picriticglasses from Hawaii, Nature, 353, 553–556, doi:10.1038/353553a0.

Clague, D. A., J. B. Paduan, W. C. McIntosh, B. L. Cousens,A. S. Davis, and J. R. Reynolds (2006), A submarine per-spective of the Honolulu Volcanics, O’ahu, J. Volcanol.Geotherm. Res., 151, 279–307, doi:10.1016/j.jvolgeores.2005.07.036.

Coltorti, M., C. Bonadiman, R. W. Hinton, F. Siena, and B. G.J. Upton (1999), Carbonatite metasomatism of the oceanicupper mantle: Evidence from clinopyroxenes and glasses inultramafic xenoliths of Grande Comore, Indian Ocean, J.Petrol., 40, 133–165, doi:10.1093/petrology/40.1.133.

Cousens, B. L. (1996), Magmatic evolution of Quaternarymafic magmas at Long Valley caldera and the Devil’s Post-pile. California: Effects of crustal contamination on litho-spheric mantle-derived magmas, J. Geophys. Res., 101,27,673–27,689, doi:10.1029/96JB02093.

Dalton, J. A., and D. C. Presnall (1998), Carbonatitic meltsalong the solidus of model lherzolite in the system CaO-MgO-Al2O3-SiO2-CO2 from 3 to 7 GPa, Contrib. Mineral.Petrol., 131, 123–135, doi:10.1007/s004100050383.

Danyushevsky, L. V., A. V. Sobolev, A. J. Crawford, M. Carroll,and R. C. Price (1993), The H2O content of basalt glasses fromSouthwest Pacific back-arc basins, Earth Planet. Sci. Lett.,117, 347–362, doi:10.1016/0012-821X(93)90089-R.

Dasgupta, R., M. M. Hirschmann, and A. C. Withers (2004),Deep global cycling of carbon constrained by the solidus ofanhydrous, carbonated eclogite under upper mantle condi-tions, Earth Planet. Sci. Lett., 227, 73–85, doi:10.1016/j.epsl.2004.08.004.

Dasgupta, R., M. M. Hirschmann, and N. Dellas (2005), Theeffect of bulk composition on the solidus of carbonated eclo-gite from partial melting experiments at 3 GPa, Contrib.Mineral. Petrol., 149, 288–305, doi:10.1007/s00410-004-0649-0.

Dasgupta, R., M. M. Hirschmann, and K. Stalker (2006), Im-miscible transition from carbonate-rich to silicate-rich meltsin the 3 GPa melting interval of eclogite + CO2 and genesisof silica-undersaturated ocean island lavas, J. Petrol., 47,647–671, doi:10.1093/petrology/egi088.

Dasgupta, R., M. M. Hirschmann, and N. D. Smith (2007),Partial melting experiments of peridotite + CO2 at 3 GPa andgenesis of alkalic ocean island basalts, J. Petrol., 48, 2093–2124, doi:10.1093/petrology/egm053.

Dautria, J. M., C. Dupuy, D. Takherist, and J. Dostal (1992),Carbonate metasomatism in the lithospheric mantle: perido-titic xenoliths from a melilititic district of the Sahara basin,Contrib. Mineral. Petrol., 111, 37–52, doi:10.1007/BF00296576.

Davis, A. S., D. A. Clague, and W. B. Friesen (1994), Petrol-ogy and mineral chemistry of basalt from the EscanabaTrough, southern Gorda Ridge, U. S. Geol. Surv. Bull.,2022, 153–170.

Dawson, J. B. (1989), Sodium carbonatite extrusions fromOldoinyo Lengai, Tanzania: Implications for carbonatitecomplex geneses, in Carbonatites: Genesis and Evolution,edited by K. Bell, pp. 255–277, Unwin Hyman, London.

Dawson, J. B., M. S. Garson, and B. Roberts (1987), Alteredformer alkalic carbonatite lava from Oldoinyo Lengai,



30 of 34

Tanzania: Inferences for calcite carbonatite lavas, Geology,15, 765 – 768, doi:10.1130/0091-7613(1987)15<765:AFACLF>2.0.CO;2.

Dixon, J. E. (1997), Degassing of alkalic basalts, Am. Mineral.,82, 368–378.

Dixon, J. E., and D. A. Clague (2001), Volatiles in basalticglasses from Loihi seamount, Hawaii: Evidence for a rela-tively dry plume component, J. Petrol., 42, 627–654,doi:10.1093/petrology/42.3.627.

Dixon, J. E., and V. Pan (1995), Determination of the molarabsorptivity of dissolved carbonate in basalitic glass, Am.Mineral., 80, 1339–1342.

Dixon, J. E., D. A. Clague, P. Wallace, and R. Poreda (1997),Volatiles in alkalic basalts from the North Arch VolcanicField, Hawaii: Extensive degassing of deep submarine-erupted alkalic series lavas, J. Petrol., 38, 911–939,doi:10.1093/petrology/38.7.911.

Dixon, J. E., L. Leist, C. Langmuir, and J.-G. Schilling (2002),Recycled dehydrated lithosphere observed in plume-influ-enced mid-ocean-ridge basalt, Nature, 420, 385–389,doi:10.1038/nature01215.

Eggler, D. H. (1978), The effect of CO2 upon partial melting ofperidotite in the system Na2O-CaO-Al2O3-MgO-SiO2-CO2

to 35 kb, with an analysis of melting in a peridotite-H2O-CO2-System, Am. J. Sci., 278, 305–343.

Eggler, D. H., and B. O. Mysen (1976), The role of CO2 in thegenesis of olivine melilitite, discussion, Contrib. Mineral.Petrol., 55, 231–236, doi:10.1007/BF00372229.

Falloon, P. J., and D. H. Green (1990), Solidus of carbonatedfertile peridotite under fluid saturated conditions, Geology,18 , 195 – 199, doi :10.1130/0091-7613(1990)018<0195:SOCFPU>2.3.CO;2.

Feigenson, M. D. (1984), Geochemistry of Kauai volcanicsand a mixing model for the origin of Hawaiian alkalic ba-salts, Contrib. Mineral. Petrol., 87, 109–119, doi:10.1007/BF00376217.

Fekiacova, Z., W. Abouchami, S. J. G. Galer, M. O. Garcia,and A. W. Hofmann (2007), Origin and temporal evolutionof Ko’olau Volcano, Hawai’i: Inferences from isotope dataon the Ko’olau Scientific Drilling Project (KSDP), the Hon-olulu Volcanics and ODP Site 843, Earth Planet. Sci. Lett.,261, 65–83, doi:10.1016/j.epsl.2007.06.005.

Fontignie, D., and J.-G. Schilling (1991), 87Sr/86Sr and REEvariations along the Easter Microplate boundaries (SouthPacific): Application of multivariate statistical analyses toridge segmentation, Chem. Geol., 89, 209 – 241,doi:10.1016/0009-2541(91)90018-M.

Frey, F. A., M. O. Garcia, and M. F. Roden (1994), Geochem-ical characteristics of Koolau Volcano: Implications of inter-shield geochemical differences among Hawaiian volcanoes,Geochim. Cosmochim. Acta, 58, 1441–1462, doi:10.1016/0016-7037(94)90548-7.

Frey, F. A., D. Clague, J. J. Mahoney, and J. M. Sinton (2000),Volcanism at the edge of the Hawaiian plume: petrogenesisof submarine alkalic lavas from the North Arch VolcanicField, J. Petrol., 41, 667–691, doi:10.1093/petrology/41.5.667.

Frey, F. A., S. Huang, J. Blichert-Toft, M. Regelous, andM. Boyet (2005), Origin of depleted components in basaltrelated to the Hawaiian hot spot: Evidence from isotopicand incompatible element ratios, Geochem. Geophys. Geo-syst., 6, Q02L07, doi:10.1029/2004GC000757.

Frezzotti, M. L., J. L. R. Touret, and E.-R. Neuman (2002),Ephemeral carbonate melts in the upper mantle: carbonate-silicate immiscibility in microveins and inclusions withinspinel peridotite xenoliths, La Gomera Canary Islands, Eur.

J. Mineral., 14, 891–904, doi:10.1127/0935-1221/2002/0014-0891.

Gaffney, A. M., B. K. Nelson, and J. Blichert-Toft (2004),Geochemical constraints on the role of oceanic lithospherein intra-volcano heterogeneity at West Maui, Hawaii, J. Pet-rol., 45, 1663–1687, doi:10.1093/petrology/egh029.

Garcia, M. O., F. A. Frey, and D. G. Grooms (1986), Petrologyof volcanic rocks from Kaula Island, Hawaii: Implicationsfor the origin of Hawaiian phonolites, Contrib. Mineral.Petrol., 94, 461–471, doi:10.1007/BF00376339.

Hammouda, T. (2003), High-pressure melting of carbonatedeclogite and experimental constraints on carbon recyclingand storage in the mantle, Earth Planet. Sci. Lett., 214,357–368, doi:10.1016/S0012-821X(03)00361-3.

Hanyu, T., D. A. Clague, I. Kaneoka, T. J. Dunai, and G. R.Davies (2005), Noble gas systematics of submarine alkaliclavas near the Hawaiian hotspot, Chem. Geol., 214, 135–155, doi:10.1016/j.chemgeo.2004.08.051.

Hart, S. R., and G. A. Gaetani (2006), Mantle Pb paradoxes:The sulfide solution, Contrib. Mineral. Petrol., 152, 295–308, doi:10.1007/s00410-006-0108-1.

Hauri, E. H., N. Shimizu, J. J. Dieu, and S. R. Hart (1993),Evidence for hotspot-related carbonatite metasomatism inthe oceanic upper mantle, Nature, 365, 221 – 227,doi:10.1038/365221a0.

Hirose, K. (1997), Partial melt compositions of carbonatedperidotite at 3 GPa and role of CO2 in alkali-basalt magmageneration, Geophys. Res. Lett., 24, 2837 – 2840,doi:10.1029/97GL02956.

Hoernle, K., G. Tilton, M. J. Le Bas, S. Duggen, and D. G.Schonberg (2002), Geochemistry of oceanic carbonatitescompared with continental carbonatites: Mantle recyclingof oceanic crustal carbonate, Contrib. Mineral. Petrol.,142, 520–542.

Hofmann, A. W., and K. P. Jochum (1996), Source character-istics derived from very incompatible trace elements in Mau-na Loa and Mauna Kea basalts, Hawaii Scientific DrillingProject, J. Geophys. Res., 101, 11,831–11,839, doi:10.1029/95JB03701.

Honda, M., I. McDougall, D. B. Patterson, A. Doulgeris, andD. A. Clague (1993), Noble gases in submarine pillow basaltglasses from Loihi and Kilauea, Hawaii: A solar componentin the Earth, Geochim. Cosmochim. Acta, 57, 859–874,doi:10.1016/0016-7037(93)90174-U.

Hou, Z., S. Tian, Z. Yuan, Y. Xie, S. Yin, L. Yi, H. Fei, andZ. Yang (2006), The Himalayan collision zone carbonatitesin western Sichuan, SW China: Petrogenesis, mantle sourceand tectonic implication, Earth Planet. Sci. Lett., 244,234–250, doi:10.1016/j.epsl.2006.01.052.

Huang, S., and F. A. Frey (2003), Trace element abundances ofMauna Kea basalt from Phase 2 of the Hawaiian ScientificDrilling Project: petrogenetic implications of correlationswith major element content and isotopic ratios, Geochem.Geophys. Geosyst., 4(6), 8711, doi:10.1029/2002GC000322.

Huang, S., and F. Frey (2005), Recycled oceanic crust in theHawaiian plume: evidence from temporal geochemical varia-tions within theKoolau Shield,Contrib. Mineral. Petrol., 149,556–575, doi:10.1007/s00410-005-0664-9.

Huang, S., F. A. Frey, J. Blichert-Toft, R. V. Fodor, G. R.Bauer, and G. Xu (2005), Enriched components in the Ha-waiian plume: Evidence from Kahoolawe Volcano, Hawaii,Geochem. Geophys. Geosyst., 6, Q11006, doi:10.1029/2005GC001012.

Ionov, D. A., C. Dupuy, S. Y. O’Reilly, M. G. Kopylova, andY. S. Genshaft (1993), Carbonated peridotite xenoliths from



31 of 34

Spitsbergen: implications for trace element signature of man-tle carbonate metasomatism, Earth Planet. Sci. Lett., 119,283–297, doi:10.1016/0012-821X(93)90139-Z.

Izraeli, E. S., J. W. Harris, and O. Navon (2001), Brine inclu-sions in diamonds: A new upper mantle fluid, Earth Planet.Sci . Le t t . , 187 , 323 – 332, doi :10.1016/S0012-821X(01)00291-6.

Jackson, L., F. W. Brown, and S. T. Neil (1987), Major andminor elements requiring individual determination, classicalwhole-rock analysis, and rapid rock analysis, in Methods ofGeochemical Analysis, edited by P. A. Baedeker, U. S. Geol.Surv. Bull., 1770, G1–G23.

Jackson, M. C., F. A. Frey, M. O. Garcia, and R. A. Wilmoth(1999), Geology and geochemistry of basaltic lava flows anddikes from the Trans-Ko’olau Tunnel, Oahu, Hawaii, Bull.Volcanol., 60, 381–401, doi:10.1007/s004450050239.

Jambon, A., B. Deruelle, G. Dreibus, and F. Pineau (1995),Chlorine and bromine abundance in MORB: the contrastingbehaviour of the Mid-Atlantic Ridge and East Pacific Riseand implications for chlorine geodynamic cycle, Chem.Geol., 126, 101–117, doi:10.1016/0009-2541(95)00112-4.

Kamenetsky, V. S., A. J. Crawford, S. Eggins, and R. Muhe(1997), Phenocryst and melt inclusion chemistry of near-axisseamounts, Valu Fa Ridge, Lau Basin: Insight into mantlewedge melting and the addition of subduction components,Earth Planet. Sci. Lett., 151, 205–223, doi:10.1016/S0012-821X(97)81849-3.

Kaneoka, I., T. Hanyu, J. Yamamoto, and Y. N. Miura (2002),Noble gas systematics of the Hawaiian volcanoes based onthe analysis of Loihi, Kilauea and Koolau submarine rocks,in Hawaiian Volcanoes, Deep Underwater Perspectives,Geophys. Monogr. Ser., vol. 128, edited by E. Takahashi etal., pp. 373–389, AGU, Washington, D. C.

Kent, A. J. R., D. A. Clague, M. Honda, E. M. Stolper, I. D.Hutcheon, and M. C. Norman (1999), Widespread assimila-tion of a seawater-derived component at Loihi Seamount,Hawaii, Geochim. Cosmochim. Acta, 63, 2749–2761,doi:10.1016/S0016-7037(99)00215-X.

Keppler, H. (2003), Water solubility in carbonatite melts, Am.Mineral., 88, 1822–1824.

Kerrick, D. M., and J. A. D. Connolly (2001), Metamorphicdevolatilization of subducted oceanic metabasalts: implica-tions for seismicity, arc magmatism and volatile recycling,Earth Planet. Sci. Lett., 189, 19–29, doi:10.1016/S0012-821X(01)00347-8.

Keshav, S., and G. Sen (2003), A rare composite xenolith fromSalt Lake Crater, Oahu: High-pressure fractionation and im-plications for kimberlitic melts in the Hawaiian mantle, Con-trib. Mineral. Petrol., 144, 548–558.

Kingsley, R. H. (2002), Mantle plume-spreading ridge interac-tion in the southeast Pacific: Isotope and trace element geo-chemistry of the Easter-Salas y Gomez Seamount chain andthe Easter Microplate, Ph.D. dissertation, Univ. of R. H.,Narragansett.

Kingsley, R. H., J. Blichert-Toft, D. Fontignie, and J.-G. Schil-ling (2007), Hafnium, neodymium, and strontium isotopeand parent-daughter element systematics in basalts fromthe plume-ridge interaction system of the Salas y GomezSeamount Chain and Easter Microplate, Geochem. Geophys.Geosyst., 8, Q04005, doi:10.1029/2006GC001401.

Klein-BenDavid, O., E. S. Izraeli, E. Hauri, and O. Navon(2004), Mantle-fluid evolution – a tale of one diamond,Lithos, 77, 243–253, doi:10.1016/j.lithos.2004.04.003.

Klein-BenDavid, O., E. S. Izraeli, E. Hauri, and O. Navon(2007), Fluid inclusions in diamonds from the Diavik mine,

Canada and the evolution of diamond-forming fluids, Geo-chim. Cosmochim. Acta, 71, 723–744, doi:10.1016/j.gca.2006.10.008.

Knaack, C., S. Cornelius, and P. R. Hooper (1994), Traceelement analysis of rocks and minerals by ICP-MS, openfile report, Dept. of Geol., Wash. State Univ., Pullman.

Kogarko, L. N., C. M. B. Henderson, and H. Pacheco (1995),Primary Ca-rich carbonatite magma and carbonate-silicate-sulphide liquid immiscibility in upper mantle, Contrib.Mineral. Petrol., 121, 267–274, doi:10.1007/BF02688242.

Kohlstedt, D. L., H. Keppler, and D. C. Rubie (1996), Solu-bility of water in the a, b, and g phases of (Mg, Fe)2SiO4,Contrib. Mineral. Petrol., 123, 345–357, doi:10.1007/s004100050161.

Kurz, M. D., W. J. Jenkins, S. R. Hart, and D. Clague (1983),Helium isotopic variations in volcanic rocks from Loihi Sea-mount and the Island of Hawaii, Earth Planet. Sci. Lett., 66,388–406, doi:10.1016/0012-821X(83)90154-1.

Kurz, M. D., M. O. Garcia, F. A. Frey, and P. A. O’Brian(1987), Temporal helium isotopic variations within Hawaiianvolcanoes: basalts from Mauna Loa and Haleakala, Geo-chim. Cosmochim. Acta, 51, 2905–2914, doi:10.1016/0016-7037(87)90366-8.

Lassiter, J. C., and E. H. Hauri (1998), Osmium-iosotope var-iations in Hawaiian lavas: Evidence for recycled oceaniclithosphere in the Hawaiian plume, Earth Planet. Sci. Lett.,164, 483–496, doi:10.1016/S0012-821X(98)00240-4.

Le Roex, A. P., and R. Lanyon (1998), Isotope and trace ele-ment geochemistry of Cretaceous Damaraland lamprophyresand carbonatites, northwestern Namibia: Evidence forplume-lithosphere interactions, J. Petrol., 39, 1117–1146,doi:10.1093/petrology/39.6.1117.

Lipman, P. W., D. A. Clague, J. G. Moore, and R. T. Holcomb(1989), South Arch volcanic field - newly identified younglava flows on the sea floor south of the Hawaiian Ridge,Geology, 17, 611–614, doi:10.1130/0091-7613(1989)017<0611:SAVFNI>2.3.CO;2.

Maaløe, S., D. James, P. Smedley, S. Petersen, and L. B.Garmann (1992), The Koloa volcanic suite of Kauai, Hawaii,J. Petrol., 33, 761–784.

Menzies, M. A., and C. J. Hawkesworth (1987), Mantle Me-tasomatism, Academic, London.

Michael, P. J., and W. C. Cornell (1998), Influence of spread-ing rate and magma supply on crystallization and assimila-tion beneath mid-ocean ridges: evidence from chlorine andmajor element chemistry of mid-ocean ridge basalts, J. Geo-phys. Res., 103, 18,325–18,356, doi:10.1029/98JB00791.

Michael, P. J., and J.-G. Schilling (1989), Chlorine in mid-ocean ridge magmas: evidence for assimilation of sea-water-influenced components, Geochim. Cosmochim. Acta,53, 3131–3143, doi:10.1016/0016-7037(89)90094-X.

Minarik, W. G., and E. B. Watson (1995), Interconnectivity ofcarbonate melt at low melt fraction, Earth Planet. Sci. Lett.,133, 423–437, doi:10.1016/0012-821X(95)00085-Q.

Molina, J. F., and S. Poli (2000), Carbonate stability and fluidcomposition in subducted oceanic crust: an experimentalstudy on H2O-CO2-bearing basalts, Earth Planet. Sci. Lett.,176, 295–310, doi:10.1016/S0012-821X(00)00021-2.

Moore, J. G., and D. A. Clague (2004), Hawaiian submarinemanganese– iron oxide crusts—a dating tool?, Geol. Soc.Am. Bull., 116, 337–347, doi:10.1130/B25304.1.

Navon, O., I. D. Hutcheon, G. R. Rossman, andG. L.Wasserburg(1988), Mantle-derived fluids in diamond micro-inclusions,Nature, 335, 784–789, doi:10.1038/335784a0.

Neumann, E.-R., and E. Wulff-Pedersen (1997), The origin ofhighly silicic glass in mantle xenoliths from the Canary



32 of 34

Islands, J. Petrol., 38, 1513–1539, doi:10.1093/petrology/38.11.1513.

Newman, S., E. Stolper, and R. J. Stern (2000), H2O and CO2

in magmas from the Mariana arc and back arc systemsGeo-chem. Geophys. Geosyst., 1(5), 1013, doi:10.1029/1999GC000027.

Ozawa, A., T. Tagami, and M. O. Garcia (2005), Unspiked K-Ar dating of the Honolulu rejuvenated and Ko’olau shieldvolcanism on O’ahu, Hawaii, Earth Planet. Sci. Lett., 232,1–11, doi:10.1016/j.epsl.2005.01.021.

Perchuk, L. L., O. G. Safonov, V. O. Yapaskurt, and J. M.Barton (2002), Crystal-melt equilibria involving potassium-bearing clinopyroxene as indicator of mantle-derived ultra-high-potassic liquids: An analytical review, Lithos, 60, 89–111, doi:10.1016/S0024-4937(01)00072-X.

Plank, T., and C. H. Langmuir (1992), Effects of the meltingregime on the composition of oceanic crust, J. Geophys.Res., 97, 19,749–19,770, doi:10.1029/92JB01769.

Presnall, D. C., and G. H. Gudfinnsson (2005), Carbonate-richmelts in the oceanic low-velocity zone and deep mantle, inPlates, Plumes & Paradigms, edited by G. H. Foulger et al.,Geol. Soc. Am. Spec. Pap., 388, 207–216.

Raga, S., R. M. Davies, W. L. Griffin, S. Jackson, and S. Y.O’Reilly (2003), Trace element analysis of diamond byLAM ICPMS: Preliminary results, paper presented at 8thInternational Kimberlite Conference, BHP Billiton, Victoria,British Columbia, Canada.

Reiners, P. W. (2002), Temporal-compositional trends in intra-plate basalt eruptions: Implications for mantle heterogeneityand melting processes, Geochem. Geophys. Geosyst., 3(2),1011, doi:10.1029/2001GC000250.

Reiners, P. W., and B. K. Nelson (1998), Temporal-composi-tional-isotopic trends in rejuvenated-stage magmas of Kauai,Hawaii, and implications for mantle melting processes, Geo-chim. Cosmochim. Acta, 62, 2347–2368, doi:10.1016/S0016-7037(98)00141-0.

Revenaugh, J., and S. Sipkin (1994), Seismic evidence forsilicate melt atop the 410-km mantle discontinuity, Nature,369, 474–476, doi:10.1038/369474a0.

Ribe, N. M., and U. R. Christensen (1999), The dynamicalorigin of Hawaiian volcanism, Earth Planet. Sci. Lett.,171, 517–531, doi:10.1016/S0012-821X(99)00179-X.

Rison, W., and H. Craig (1983), Helium isotopes and mantlevolatiles in Loihi Seamount and Hawaiian Island basalts andxenoliths, Earth Planet. Sci. Lett., 66, 407 – 426,doi:10.1016/0012-821X(83)90155-3.

Roden, M. F., F. A. Frey, and D. A. Clague (1984), Geochem-istry of tholeiitic and alkalic lavas from the Koolau Range,Oahu, Hawaii: Implications for Hawaiian volcanism, EarthPlanet. Sci. Lett., 69, 141 – 158, doi:10.1016/0012-821X(84)90079-7.

Roden, M. F., T. Trull, S. R. Hart, and F. A. Frey (1994), NewHe, Sr, Nd and Pb isotopic constraints on the constitution ofthe Hawaiian plume: results from Koolau Volcano, Oahu,Hawaii, Geochim. Cosmochim. Acta, 58, 1431–1440,doi:10.1016/0016-7037(94)90547-9.

Rudnick, R. L., W. F. McDonough, and B. W. Chappell(1993), Carbonatite metasomatism in the northern Tanzanianmantle: petrographic and geochemical characteristics, EarthPlanet. Sci. Lett., 114, 463–476, doi:10.1016/0012-821X(93)90076-L.

Rupke, L., J. P. Morgan, and J. E. Dixon (2006), Implicationsof subduction rehydration for the Earth’s deep water cycle, inEarth’s Deep Water Cycle, Geophys. Monogr. Ser., vol. 168,edited by S. D. Jacobsen and S. van der Lee, pp. 263–276,AGU, Washington, D. C.

Salters, V. J. M., and A. Zindler (1995), Extreme 176Hf/177Hfin the sub-oceanic mantle, Earth Planet. Sci. Lett., 129, 13–30, doi:10.1016/0012-821X(94)00234-P.

Salters, V. J. M., J. E. Longhi, and M. Bizimis (2002), Nearmantle solidus trace element partitioning at pressures up to3.4 GPa, Geochem. Geophys. Geosyst., 3(7), 1038,doi:10.1029/2001GC000148.

Salters, V. J. M., J. Blichert-Toft, Z. Fekiacova, A. Sachi-Kocher,and M. Bizimis (2006), Isotope and trace element evidence fordepleted lithosphere in the source of enriched Ko’olau basalts,Contrib. Mineral. Petrol., 151, 297–312, doi:10.1007/s00410-005-0059-y.

Schiano, P., and R. Clocchiatti (1994), Worldwide occurrenceof silica-rich melts in sub-continental and sub-oceanic man-tle minerals, Nature, 368, 621–624, doi:10.1038/368621a0.

Schiano, P., R. Clocchiatti, N. Shimizu, D. Weis, and N. Mat-tielli (1994), Cogenetic silica-rich and carbonate-rich meltstrapped in mantle minerals in Kerguelen ultramafic xeno-liths: Implications for metasomatism in the oceanic uppermantle, Earth Planet. Sci. Lett., 123, 167 – 178,doi:10.1016/0012-821X(94)90265-8.

Schilling, J.-G., R. Kingsley, D. Fontignie, R. Poreda, andS. Xue (1999), Dispersion of the Jan Mayen and Icelandmantle plumes in the Arctic: A He-Pb-Nd-Sr isotope tracerstudy of basalts from the Kolbeinsey, Mohns, and Knipo-vich Ridges, J. Geophys. Res., 104, 10,543–10,569,doi:10.1029/1999JB900057.

Schrauder, M., and O. Navon (1994), Hydrous and carbonatiticmantle fluids in fibrous diamonds from Jwaneng, Botswana,Geochim. Cosmochim. Acta, 58, 761–771, doi:10.1016/0016-7037(94)90504-5.

Schrauder, M., C. Koeberl, and O. Navon (1996), Trace ele-ment analyses of fluid-bearing diamonds from Jwaneng,Botswana, Geochim. Cosmochim. Acta, 60, 4711–4724,doi:10.1016/S0016-7037(96)00274-8.

Schultz, F., B. Lehman, S. Tawackoli, R. Rossling, B. Belyatski,and P. Dulski (2004), Carbonatite diversity in the CentralAndes: The Ayopaya alkaline province, Bolivia, Contrib.Mineral. Petrol., 148, 391–408.

Sen, G., A. Macfarlane, and N. Srimal (1996), Significance ofrare hydrous alkaline melts in Hawaiian xenoliths, Contrib.Mineral . Petrol . , 122 , 415 – 427, doi :10 .1007/s004100050137.

Silva, L. C., M. J. Le Bas, and A. H. F. Robertson (1981), Anoceanic carbonatite volcano on Santiago, Cape Verde Is-lands, Nature, 294, 644–645, doi:10.1038/294644a0.

Silver, L. A., P. D. Ihinger, and E. M. Stolper (1990), Theinfluence of bulk composition on the speciation of water insilicate glasses, Contrib. Mineral. Petrol., 104, 142–162,doi:10.1007/BF00306439.

Simons, K., J. E. Dixon, J.-G. Schilling, R. Kingsley, and R.Poreda (2002), Volatiles in basaltic glasses from the Easter-Salas y Gomez Seamount Chain and Easter Microplate: Im-plications for geochemical cycling of volatile elements, Geo-chem. Geophys. Geosyst., 3(7), 1039, doi:10.1029/2001GC000173.

Sims, K. W. W., D. J. DePaolo, M. T. Murrell, W. S. Baldridge,S. J. Goldstein, and D. A. Clague (1995), Mechanism of mag-ma generation beneath Hawaii and mid-ocean ridges: ura-nium/thorium and samarium/neodymium isotopic evidence,Science, 267, 508–512, doi:10.1126/science.267.5197.508.

Sisson, T. W., and G. D. Layne (1993), H2O in basalt andbasaltic andesite glass inclusions from four subduction-relatedvolcanoes, Earth Planet. Sci. Lett., 117, 619 – 635,doi:10.1016/0012-821X(93)90107-K.



33 of 34

Sisson, T. W., P. W. Lipman, and J. Naka (2002), Submarinealkalic through tholeiitic shield-stage development of Ki-lauea Volcano, Hawaii, in Hawaiian Volcanoes: Deep Un-derwater Perspectives, Geophys. Monogr. Ser., vol. 128,edited by E. Takahashi et al., pp. 193–219, AGU,Washington, D. C.

Smyth, J. R., T. Kawamoto, S. D. Jacobsen, R. J. Swope, R. L.Hervig, and J. R. Holloway (1997), Crystal structure ofmonoclinic hydrous wadsleyite b- (Mg, Fe)2SiO4, Am.Mineral., 82, 270–275.

Song, T. A., D. B. Helmberger, and S. P. Grand (2004), Low-velocity zone atop the 410-km seismic discontinuity in thenorthwestern United States, Nature, 427, 530–533,doi:10.1038/nature02231.

Stearns, H., and G. A. Macdonald (1947), Geology andground-water resources of the Island of Niihau, Hawaii, U.S. Geological Survey, Hawaii Div. Hydrogr. Bull., 12, 51 pp.

Stille, P., D. M. Unruh, and M. Tatsumoto (1986), Pb, Sr, Nd,and Hf isotopic constraints on the origin of Hawaiian basaltsand evidence for a unique mantle source, Geochim. Cosmo-chim. Acta , 50 , 2303 – 2319, doi:10.1016/0016-7037(86)90084-0.

Stolper, E., and S. Newman (1994), The role of water in thepetrogenesis of Mariana trough magmas, Earth Planet. Sci.Lett., 121, 293–325, doi:10.1016/0012-821X(94)90074-4.

Tappe, S., S. F. Foley, G. A. Jenner, L. M. Heaman, B. A.Kjarsgaard, R. L. Romer, A. Stracke, N. Joyce, and J. Hoefs(2006), Genesis of ultramafic lamprophyres and carbonatitesat Aillik Bay, Labrador: A consequence of incipient litho-spheric thinning beneath the North Atlantic craton, J. Petrol.,47, 1261–1315, doi:10.1093/petrology/egl008.

Tappe, S., S. F. Foley, A. Stracke, R. L. Romer, B. Z. Kjarsgaard,L. M. Heaman, and N. Joyce (2007), Craton reactivation onthe Labrador Sea margins: 40Ar/39Ar age and Sr-Nd-Hf-Pbisotope constraints from alkaline and carbonatite intrusives,Earth Planet. Sci. Lett., 256, 433–454, doi:10.1016/j.epsl.2007.01.036.

Tatsumoto, M. (1978), Isotopic composition of lead in oceanicbasalts and its implications to mantle evolution, Earth Planet.Sci. Lett., 38, 63–87, doi:10.1016/0012-821X(78)90126-7.

ten Brink, U. (1991), Volcano spacing and plate rigidity, Geol-ogy, 19, 397–400, doi:10.1130/0091-7613(1991)019<0397:VSAPR>2.3.CO;2.

Todt, W., R. A. Cliff, A. Hanser, and A. W. Hofmann (1984),202Pb-205Pb spike for Pb isotopic analysis, Terra Cognita,4, 209.

Valbracht, P. J., H. Staudigel, M. Honda, I. McDougall, andG. R. Davies (1996), Isotopic tracing of volcanic sourceregions from Hawaii: decoupling of gaseous from lithophilemagma components, Earth Planet. Sci. Lett., 144, 185–198,doi:10.1016/0012-821X(96)00126-4.

Vannucci, R., P. Bottazzi, E.Wulff-Pedersen, and E.-R. Neumann(1998), Partitioning of REE, Y, Sr, Zr and Ti between clinopyr-oxene and silicate melts in the mantle under La Palma (CanaryIsland): implications for the nature of the metasomatic agents,Earth Planet. Sci. Lett., 158, 39–51, doi:10.1016/S0012-821X(98)00040-5.

Vinnik, L., and V. Farra (2002), Subcratonic low-velocity layerand flood basalts, Geophys. Res. Lett., 29(9), 1049,doi:10.1029/2001GL014064.

Walker, J. A., K. Roggensack, L. C. Patino, B. I. Cameron, andM. Otoniel (2003), The water and trace element contents ofmelt inclusions across an active subduction zone, Contrib.Mineral. Petrol., 146, 62–77, doi:10.1007/s00410-003-0482-x.

Wallace, P. J. (1998), Water and partial melting in mantleplumes: Inferences from the dissolved H2O concentrationsof Hawaiian basaltic magmas, Geophys. Res. Lett., 25,3639–3642, doi:10.1029/98GL02805.

Wirth, R., and A. Rocholl (2003), Nanocrystalline diamondfrom the Earth’s mantle underneath Hawaii, Earth Planet.Sci. Lett., 211, 357–369, doi:10.1016/S0012-821X(03)00204-8.

Woolley, A. R., and D. R. C. Kempe (1989), Carbonatites:Nomenclature, average chemical compositions, and elementdistribution, in Carbonatites: Genesis and Evolution, editedby K. Bell, pp. 1–114, Unwin Hyman, London.

Workman, R. K., and S. R. Hart (2005), Major and trace ele-ment composition of the depleted MORB mantle (DMM),Earth Planet. Sci. Lett., 231, 53–72.

Wulff-Pedersen, E., E.-R. Neumann, and B. B. Jensen (1996),The upper mantle under La Palma, Canary Islands: formationof Si-K-Na-rich melt and its importance as a metasomaticagent, Contrib. Mineral. Petrol., 125, 113 – 139,doi:10.1007/s004100050210.

Wulff-Pedersen, E., E.-R. Neumann, R. Vannucci, P. Bottazzi,and L. Ottolini (1999), Silicic melts produced by reactionbetween peridotite and infiltrating basaltic melts: ion probedata on glasses and minerals in veined xenoliths from LaPalma, Canary Islands, Contrib. Mineral. Petrol., 137, 59–82, doi:10.1007/s004100050582.

Wyllie, P. J. (1988), Solidus curves, mantle plumes, and mag-ma generation beneath Hawaii, J. Geophys. Res., 93, 4171–4181, doi:10.1029/JB093iB05p04171.

Wyllie, P. J. (1989), Origin of carbonatites: Evidence fromphase equilibrium studies, in Carbonatites: Genesis andEvolution, edited by K. Bell, pp. 500–560, Unwin Hyman,London.

Wyllie, P. J., and I. D. Ryabchikov (2000), Volatile compo-nents, magmas, and critical fluids in upwelling mantle,J. Petrol., 41, 1195–1206, doi:10.1093/petrology/41.7.1195.

Xu, G., F. A. Frey, D. A. Clague, D. Weis, and M. H. Beeson(2005), East Molokai and other Kea-trend volcanoes: Mag-matic processes and sources as they migrate away from theHawaiian hot spot, Geochem. Geophys. Geosyst., 6, Q05008,doi:10.1029/2004GC000830.

Yang, H.-J., F. A. Frey, and D. A. Clague (2003), Constraintson the source components of lavas forming the HawaiianNorth Arch and Honolulu Volcanics, J. Petrol., 44, 603–627, doi:10.1093/petrology/44.4.603.

Yaxley, G. M., and G. P. Brey (2004), Phase relations ofcarbonate-bearing eclogite assemblages from 2.5 to 5.5 GPa:implications for petrogenesis of carbonatites,Contrib. Mineral.Petrol., 146, 606–619, doi:10.1007/s00410-003-0517-3.

Yaxley, G. M., and D. H. Green (1994), Experimental demon-stration of refractory carbonate-bearing eclogite and siliceousmelt in the subduction regime, Earth Planet. Sci. Lett., 128,313–325, doi:10.1016/0012-821X(94)90153-8.

Yaxley, G. M., D. H. Green, and V. Kamenetsky (1998), Car-bonatite metasomatism in the southeastern Australian litho-sphere, J. Petrol., 39, 1917–1930, doi:10.1093/petrology/39.11.1917.



34 of 34

Carbonatite and silicate melt metasomatism of depleted mantle surrounding the Hawaiian plume: origin...

Documents

Transcript of Carbonatite and silicate melt metasomatism of depleted mantle surrounding the Hawaiian plume: origin...